2022

ISSN 1015-5589, Vol. XXXI, No.. 1

World Safety Journal

© Copyright 2022 | WSO

A Multi-Industry Analysis of Human-Machine Systems: The Connection to Truck

In This Edition

Automation
by Todd Pascoe, Shirley McGough, and Janis Jansz

WSO’s Certification Program is accredited, in

Analysis of Australian Coronial Inquest and Non-Inquest findings of heavy compliance with the ISO/IEC 17024:2012
vehicle fatal crashes reveals shortcomings in investigations
by Ivan Cikara, Geoff Dell, and Aldo Raineri

Fast-track Industry Safety Culture and Reduction of Incidents/Losses

by Harbans Lal

Public Transportation Infrastructure Projects and Financial Risks

by Elias M. Choueiri
World Safety Organization

Statement of Purpose
and Objectives
WSO’s purpose is to internationalize all
safety fields, including occupational and
environmental safety and health, accident
prevention movement, etc., and to
disseminate throughout the world the
practices skills, arts, and technologies of
safety and accident prevention.

WSO’s objective is to protect people,

property, resources, and the environment on
local, regional, national, and international
levels. WSO membership is open to all
individuals and entities involved in the
safety and accident prevention field,
regardless of race, color, creed, ideology,
religion, social status, sex, or political
beliefs.

WSO is in Consultative Category II Status

(Non-Governmental Organization - NGO)
to the Economic and Social Council of the
United Nations.

WSO is a Not-for-Profit Corporation

(Missouri, USA), non-sectarian, non-
political movement dedicated to:

“Making Safety a Way of Life … Worldwide.”

 . Table of Contents
A Multi-Industry Analysis of Human-Machine Systems: The Connection to Truck
Automation
by Todd Pascoe, Shirley McGough, and Janis Jansz ..………………………….... Pages 1-38

Analysis of Australian Coronial Inquest and Non-Inquest findings of heavy vehicle

fatal crashes reveals shortcomings in investigations
by Ivan Cikara, Geoff Dell, and Aldo Raineri ………………………………………. Pages 39-51
Editor-in-Chief
Fast-track Industry Safety Culture and Reduction of Incidents/Losses
Prof. Dr. Elias M. Choueiri by Harbans Lal …………………………………………………………………………….…… Pages 52-58
Member of the WSO Board of Directors
and WSO Liaison Officer to the United Nations Public Transportation Infrastructure Projects and Financial Risks
by Elias M. Choueiri …………………………………………………………………………. Pages 59-70
Associate Editor
Dr. Janis K. Jansz Article Submission
Curtin University, Western Australia
The World Safety Journal (WSJ) is a quarterly refereed journal (ISSN: 1015-5589) that
Director of the WSO National Office for Australia, and includes original full-length papers, review articles and case studies on the most recent
Member of the WSO Board of Directors outcomes of advanced research in any occupational safety and health fields.
All articles shall be written in concise English and typed with a minimum font size of 12
Editorial Board point. Articles should have an abstract of not more than 300 words. Articles shall be
submitted as Times New Roman print and presented in the form the writer wants
Dr. David P. Gilkey published. On a separate page, the author should supply the author’s name, contact
details, professional qualifications, current employment position, a brief bio, and a photo
Montana Tech University, USA
of the author. This should be submitted with the article.
Dr. Richard Franklin Writers should include all references and acknowledgments. Authors are responsible for
James Cook University, Queensland, Australia ensuring that their works do not infringe on any copyright. Failure to do so can result in
the writer being accountable for breach of copyright. The accuracy of the references is the
Dr. Charles H. Baker author’s responsibility.
World Safety Organization, USA

Dr. Milos Nedved References

Central Queensland University, Australia
Articles should be referenced according to the Publication Manual of the American
Dr. Michael Thomas Psychological Association ,7th ed.
Department of Transportation (FAA), USA Books are referenced as follows:
Author. (Year of publication). Title of publication. Publisher.
Cover Photo Articles are referenced as follows:
Author (Year). Title of article. Name of Journal. Volume (Issue), Page numbers of article.
SKYWAY Technology
Internet information is referenced as follows:
SkyWay is a pole-supported rail transport overpass Name of author. (Year of publication). Title. DOI or web address if no DOI.
designed to transfer special rolling stock equipped with
an anti-derailment system. In itself, the rail SkyWay car
is a kind of a conventional car mounted on steel wheels
Submissions should be mailed to:
Just as a traditional car, it can be driven by a diesel
engine, a gasoline engine, a turbine or a combined drive.
If necessary, the engine can run on natural gas, methane, WSO, World Management Center
hydrogen, alcohol and other clean fuels. In addition, Attn: Editorial Staff /
electrification of the rail vehicle is possible with the use Prof. Dr. Elias M. Choueiri
of an external source of electrical energy (like a trolley World Management Center, 106 W Young St #F
bus, tram or subway). Self-generated power supply can Warrensburg, MO 64093, USA
also be used, e.g. on-board batteries, energy storage units or emailed to: [email protected]
of capacitor, molecular or any other type, fuel cells and
others. The basis of the track structure is tension Articles, wherever possible, must be up-to-date and relevant to the Safety Industry. All
prestressed uncut string rails or a tension prestressed articles are Blind Peer Reviewed by at least two referees before being accepted for
uncut string-truss structure (heavy and multi-functional publication.
transport system). String-rail track structure has no joints
and expansion (temperature) seams and is noted for a
perfect straightness and smooth curves, which is crucial
to achieve the highest operational performance of the Disclaimer
transport system – high speed, minimum impact stress,
low energy consumption. Opinions expressed by contributors in articles, or reproduced articles, are the individual
opinions of such contributors or the authors and not necessarily those of the World
Retrieved from: Safety Organization. Reproduction of articles or abstracts contained in this journal is
https://sky-way.org/en/technology-SkyWay/ approved, providing the source is acknowledged.
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 1

World Safety Journal

A peer-reviewed journal,
published by the World Safety Organization

Journal Homepage:
https://worldsafety.org/wso-world-safety-journal/

A Multi-Industry Analysis of Human-Machine Systems: The Connection to Truck

Automation
Todd Pascoe, Shirley McGough and Janis Jansz
Curtin University, Western Australia

KEYWORDS ABSTRACT
Automation This research is a critical review of safety-related themes in human-machine
Mining systems across multiple industries. The aim is to explore the lessons of engineering
Human factors human-machine systems and the residual consequences of introducing driverless
Human-machine systems trucks on a Western Australian (WA) mine site. The method involved the
Driverless trucks
identifying key words, phrases and contributing factors leading to driverless truck
events to-date. An eligibility criterion aided the selection of relevant human factors
research in the field of artificial intelligence, automated systems and augmentation.
Literature is categorised into 9 publication types, with 11 separate industries
associated within 182 pieces of material. Three broad categories were synthesised
to include: (i) technology; (ii) processes; and (iii) human factors, with three
research questions answering how this research applies to truck automation. Within
those categories, 23 research themes were found under the human-machine system
domain. The findings highlight the Mining Industry’s knowledge gaps and informs
the design of driverless technology, formation of work processes and the
accommodation of local human adaption. Conclusions provide a way forward for
the industry and pass on lessons learnt to avoid automation pitfalls.

1. INTRODUCTION

M ore than a decade ago, Rio Tinto trialed the first driverless haul truck in Western Australia
(WA). Driverless haul trucks do not need a safety driver and operate independently via
machine algorithms (Hamada & Saito, 2018). An algorithm is responsible for controlling the
actions of the haul truck, with every truck operating within the same operating parameters.
The only difference is that Mine Control gives individual truck assignments in order to deliver the daily
plan. In addition, system-based roles and ancillary equipment operators are given residual tasks to help
the driverless trucks through non-designed situations (Caterpillar, 2013). Therefore, the haul trucks are
semi-automated and interact frequently with humans in performing operational tasks.

Driverless haul trucks introduced a new set of hazards and risks, which appeared to be transforming the
risk profile of mine sites who were deploying automated technology (Department of Mines and
Petroleum, 2014b). The inherent nature of automated system design and architecture introduce

properties like complexity, reductionism, literalism, and brittleness (Billings, 2018; Dekker, 2014b; Ito


Corresponding Author: [email protected]
https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 2

& Howe, 2016). An engineered human-machine system can be considered a ‘joint’ system, where both
agents are required to collaborate as a team (Christofferson & Woods, 2002). It is evident that the
Aviation Industry has learnt the most on how to cooperate human and machine in one system. This
way, beyond isolation, the two agents can work collaboratively to become more resilient in times of
disruption. Mining companies invest in driverless technology based on the potential of making their
supply chains a lot safer and more productive (Palmer, 2019). However, despite the hype around
removing ‘driving errors’, the technology has simply removed human exposure to driving trucks and
transformed what remained (Department of Mines and Petroleum, 2015a).

The number of significant driverless truck incidents illustrates the importance of human factors
research, currently six publicly reported since 2014 in WA alone illustrates the importance of human
factors research (Department of Mines and Petroleum, 2014b; Jamasmie, 2019; McKinnon, 2019).
Such an emergence could hinder the deployment of driverless technology due to the complex nature of
unconventional incidents. Moreover, the landscape in mining operations is swiftly evolving as more
products and vendors enter the market, with human factors playing a vital role. Human factors in this
digital age is argued to be “people in systems, rather than people versus systems” (Dekker, 2019, p.
xix). Such a view will allow the Western Australian (WA) Mining Industry to become more human-
centered when designing and deploying driverless technology (Giacomin, 2015). Therefore, as a joint
human and machine system, despite being two completely different agents, they should complement
one another. Thus, within the context of human-machine systems, human factors study the design of
technology to suite the attention, memory and perceptions of humans. More specifically, taking the
study of human cognition into the ‘real world’ and understanding the interactions people have in
complex systems (Rankin et al., 2016).

As a consequence, cognitive systems engineering has progressively become popular with the expansion
of computerised systems (de Vries, 2017; Hew, 2016; Woods & Hollnagel, 2006). Researchers are
already aware of the reverberations of automated technology and the human-machine breakdowns that
have occurred across various industries. Waves of automation and technological disruption can be
identified in: Aviation, which included automated flight capabilities (Sarter, 2008); Manufacturing,
comprised of product assembly and machining (Frohm et al., 2006); Healthcare involving ICU devices
and monitoring equipment (Dominiczak & Khansa, 2018), Nuclear encompassing plant status and real-
time decision making assistance (Schmitt, 2012); Maritime including advances in communication and
navigation equipment (de Vries, 2017); Mining equipment that comprises of haul trucks and production
drills (Department of Mines and Petroleum, 2015a), and Transportation that deploys driverless cars,
trains, trucks and buses (Fridman et al., 2018; Gschwandtner et al., 2010).

Despite there being various perspectives concerning automated mining equipment (Bellamy & Pravica,
2011), it is argued that driverless haul truck safety has not been given enough attention. The full extent
of the human factors that apply in driverless haul truck systems are yet to be explored. There are in fact
perspectives that concentrate on designing remote operating equipment that is user centered (Horberry,
2012; Horberry et al., 2011), and the benefits of removing human exposure through remote control
(Fisher & Schnittger, 2012). Further perspectives argue the need for more human factors research given
safety outcomes are unknown (Lynas & Horberry, 2010), while others claim automation reduces
‘human error’ (Hamada & Saito, 2018) and increases safety through obstacle detection (Brundrett,
2014). Furthermore, there are interviews such as Lynas and Horberry (2010) that concentrate on
developers and users of technology, which explore the cognitive capacities required to operate
equipment remotely.

Undoubtedly, the literature is yet to understand how human factors research applies to haul truck
automation, an opportunity that underpinned the reason for undertaking this research. More

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 3

specifically, the review attempts to address the following questions: (1) How are the theoretical
viewpoints of human-machine systems influencing the approach to haul truck automation? (2) What
processes are designed to support automation, and do they equip human supervisors to improvise in
non-designed situations? (3) Does human adaptive behavior manage unanticipated machine
performances and the decisions to intervene or not during beyond design performances? This review
draws on human factors research from other industries that have adopted and deployed automated
technology, applying the concepts and lessons learnt to fast-forward the WA Mining Industry’s
thinking to equip them for this digital revolution.

2. METHODOLOGY
2.1 Introduction

To outline the process for identifying and organizing the research on the topic, the steps proposed by
Creswell and Creswell (2017) were embraced. The methodological process encompassed the following
steps:

1. Identify studies and key words, search databases and websites

2. Collect at least 50 research studies, prioritise them and validate the abstracts, chapters and
conclusions
3. Design a literature map to visually represent the groupings
4. Summarise and organize the literature into themes and concepts to identify opportunities

Identification Selection and Validation Analysis and Synthesis Total Visual Representation

Scientific research imported Analysis of scientific content • 96 Articles Aggregate of themes and
Published articles, books, • 22 Books phenomenon grouped from selected
into EndNote for review and to identify situations and
conference proceedings, technical
•
Scientific

validation (n = 294) themes (n = 136) 6 Reports scientific research (n = 23)

reports collected through database
• 12 Conferences
searches and reference list reviews
(n = 418)

Exclusion of generic artificial Exclusion of architecture- Exclusion of research focused on

intelligence articles and related studies not applicable to technical design, architecture and
automated systems (n = 124) human factors (n = 158) reliability (n = 0)

Documents, webpages and • 24 Webpages Aggregate of themes and

Grey Literature

Literature imported into EndNote for Analysis of literature content to

videos retrieved from websites • 8 Reports phenomenon grouped from selected
review and validation (n = 69) identify incidents, standards and
(n = 209) • 10 Videos grey literature (n = 0)
experiences (n = 46)
• 2 Standards
• 2 Presentations

Exclusion of generic webpages, Exclusion of information did not Exclusion of information that could
videos and announcements (n = 140) relate to human factors working not be related to mining operations
with automated systems (n = 23) (n = 0)

Figure 1. Process flowchart for reviewing the literature

2.2 Study and Literature Identification

The literature review originally began by evaluating the two driverless incidents reports (Department of
Mines and Petroleum, 2014b; 2015c). Both reports were analysed to identify key words, phrases and
contributing factors that led to the event. The first report (Department of Mines and Petroleum, 2014b,
pp. 1-2) provided a summary of the hazards. Through seeking the safe use of mobile autonomous
equipment, the safety bulletin identified “detection systems” and “remotely overriding” as a factor of
design in driverless haulage. Human factors included responding to “system information and
warnings”, misinterpreting “system information”, “lack of knowledge and understanding” and “not
adhering to clearance zones”. Secondly, Department of Mines and Petroleum (2015c, pp. 1-2)
summarised an incident between a manually watercart and a driverless haul truck. The report noted that

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 4

“assigning roads in the control system were inadequate”. In addition, the watercart driver was not
aware of the autonomous truck’s direction despite an in-cab awareness system to “monitor the
autonomous truck’s path”. These key words and phrases used in this report provided the basis for
searching for research studies associated with human-machine systems.

2.3 Selection and Validation

The selection of literature was based on an eligibility criterion. To be selected the literature (scientific
or grey) needed to be relevant in the fields of artificial intelligence, automated systems, or
augmentation. In addition, the literature needed to be applicable to human factors, which could then
allow similarities to be drawn in how people work with artificial agents. More importantly, the
situations where humans are successful and sometimes fail, ultimately leading or avoiding undesired
situations. Firstly, the abstracts of the research papers and introductions were evaluated based on their
intent. For example, if the literature was not designed to understand how humans and machines work
together, then it was excluded. The excluded writings were arranged into their reasons for exclusion.
Secondly, the content of the literature was evaluated for substance and relevance, excluding those that
could not be impactful in a mining context. Thirdly, the writings that were more centered around
human adaption, cognition and response were included, while technical architecture of the automated
system were removed. Despite this, a majority of scientific papers focused on the human element
working with a machine. Lastly, the literature found to be unsuitable for inclusion were used in the
introduction for context setting.

Table 1. Criteria for inclusion and exclusion

Selection Component Scientific Literature Grey Literature
Conference proceedings, peer- Government reports and standards, publicly
reviewed articles, books and chapters, released incidents, YouTube videos,
interviews. announcements, Company tutorials and
public engagements.
Inclusion Title Key words: automation, driverless, Key words: haul truck automation,
autonomous haul trucks, human driverless, autonomous haulage, haul truck
factors, augmentation, artificial incident.
intelligence.
Abstract Articles relating to the human factors
in automated systems

Content Human factors research orientated Details of reports and situations, code of
towards understanding situations, practices highlighting risks and hazards,
experiences and adaptions of humans issues with application, workplace incidents
while working with artificial systems. and anecdotal experiences
Exclusion Title Generic artificial intelligence and Generic webpages, videos and
automated system articles announcements with no correlations with
driverless/ automated haulage
Abstract Design and architecture-related
studies that did not explore associated
human factors
Content Research focused on technical design, Writings paraphrasing the intent and
architecture and network reliability purpose of driverless haulage, no specific
relation to how the technology works
practically.

2.4 Analysis and Synthesis

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 5

The selected literature was categorised into their associated publication type. The purpose of analysing
associated publication types helps frame where the research was publicised. This was necessarily given
that the technology is relatively new to the WA Mining Industry and academic research is yet to
explore. Moreover, it also highlights the magnitude of research that can be drawn from other industries
who have already deployed automated systems. As Figure 1 illustrates, a majority of literature included
in the research were scientific papers. This can be explained by the volume of research that has been
undertaken in the Aviation Industry shown in Figure 2. The significance of grey literature (i.e. web
pages, online videos) highlights the methods currently being used to understand the topic. Once
innovation tapers and competitive advantages plateau, perhaps more academic research in the field of
human-machine systems can be undertaken in the Mining Industry.

100

80

60

40

20

0
Other Presentation Government Online Video Report Conference Book Web Page Journal
Standards Proceedings Article

Figure 2. Analysis of literature by type

The timeframe and industry of focus of research were considered in Figure 3. There were 11 industries
identified within the 182 pieces of literature included in the research. By including the industry where
the research was undertaken, readers are given an indication of where automated systems have been
deployed in industry. More specifically, where research has been able to take place and explore the
consequences of replacing human work. Figure 2 illustrates how the Aviation Industry was the first
industry to explore associated human factors. From there, healthcare, manufacturing, maritime and
other associated industries have been able to leverage from those insights. The whole purpose of this
literature review is to do exactly that for the WA Mining Industry. Therefore, the industry can avoid the
pitfalls of automation, leveraging the lessons learnt from existing research and optimise their current
designs and systems of work.

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 6

Figure 3. Analysis of publication timeframe by industry type

There were three broad categories that were identified in the literature:

(i) Technology;
(ii) Processes; and
(iii) Human Factors.

These categories also contained sub-themes that provided additional context to the category. For
example, machine modes formed a part of technology, which was reinforcing how the technology
presented in the workplace. In addition, mode awareness was how well people were being made aware
of the machine modes and the complexities behind it.

By providing themes, readers are able to clearly understand the phenomenon that research has
identified thus far. Therefore, the illustration of a mind map in Figure 4 provides a visual representation
of human-machine system topics. The identified topics and associated findings can then be used for
academics, mining operators and regulators to further explore individual topics further.

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 7

Figure 4. Mind map abstraction of research themes associated with human-machine systems

3. RESULTS

3.1 Engineering Human-Machine Systems

Reductionism and Complexity

Reductionism simplifies haulage systems into their most basic parts (Dekker, 2010). The parts are
made up of driving to a load source, awaiting to be loaded by an excavator, driving to a destination and
then dumping (Caterpillar, 2013). When this process is complete, it is then repeated. Systems are
simplified in this way to enable technology to change out human tasks that it can perform (Lake et al.,
2016). Often, the replacement is dependent on what technological advancements make it viable
(Panetta, 2019). That means that driverless trucks must achieve or exceed human level performance.
Although, human level performance is evaluated in isolation, the study is only a constituent part in a
complex whole. That constituent part is then reverse engineered into a machine, following a narrow set
of instructions (Fridman et al., 2018). On the surface, the restructured haulage system can appear to
operate as intended. However, it is the reverberations along the fringes where the consequences take
place (Department of Mines and Petroluem, 2015c). A driverless truck, for instance, may be unable to
achieve its dump destination due to material being placed in the way. A human is now required to
remove that material or redirect the truck to a new dump location. This example highlights the
characteristics of the system, complex interactions between components. Therefore, the properties of
the system arise after drivers have been replaced, which can be difficult to predict (Department of
Mines and Petroluem, 2014b). Complex systems create their own individual structures, which can be
defiant of the product designer. In response to the introduction, the environment modifies and
restructures the entire system (Dekker, 2014b).

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 8

Complex systems cannot be understood through the analysis of independent tasks. Complexity evolves
through the interactions of the systems’ components (Cilliers, 2002a). Furthermore, automation creates
interdependencies that generate non-linear relationships. Therefore, systems are not linear input-output
devices as metaphorically described (McCarthy et al., 2000). Humans will not simply undertake one
task, while an automated system uninterruptedly perform another (Mitchell, 2018). A system is a
complex web of component dependencies, transformations, trade-offs and influences (Dekker, 2019).
Although automation has freed trucks drivers to perform ‘more important’ tasks, the reality is that the
truck fleet becomes silent, awkward and difficult to instruct (Christoffersen & Woods, 2002). What
used to be a mine site filled with radio calls, now quietly and independently executes the task. The
difficult and clumsy part, however, is that the apparent simplicities turn into actual complexities
(Woods, 2018). The automation of sub-driving tasks, which asks Engineers to focus on the
components. This is quite appealing when attempting to seek ways to produce and optimise at a lower
cost (Caterpillar, n.d.-a). Inefficiencies are targeted, ironing out the variability and increasing the
predictability (Hamada & Saito, 2018). However, the simplification can be achieved from what is
excluded. Complex systems are ignorant of local control and external influences that leave the system
vulnerable beyond engineering predictions (Chandler, 2014). For example, a design engineer who is
located internationally, can simply change a filter that can impact the vendors entire fleet across
Western Australia. Hence, the reason why haulage systems are now becoming more complex. The
connections are becoming wider, closely connected to a socio-technical system that cannot be isolated
(Bellamy & Pravica, 2011).

The analysis of what a system contains will not explain what it will do. The components will react
differently, depending on the type and number of influential factors (Dekker et al., 2012). The
properties will emerge once they interact in the workplace. For example, a truck is unable to identify a
wet road, therefore the interaction will require traction controls to avoid losing control (Jamasmie,
2019). Upon realising the trucks’ limitations, system supervisors will install speed restrictions on haul
routes to avoid truck slides (Department of Mines and Petroluem, 2015a). This is why the
reconstruction of systems with machine agents sometimes fail; the non-linearity of the consequences
does not represent the entire system. Despite the neat allocation of functions (de Winter & Dodou,
2011), the activities are derived from arbitrary views of human-machine strengths and weaknesses
(Dekker & Woods, 2002b). The problem is that they are never fixed, the capabilities and limitations
evolve as people learn and technological systems are upgraded (Lake et al., 2016; Woodward & Finn,
2016). Moreover, automation systems can only operate within the confines of the data they were
programmed upon (Earley, 2016). This often leaves users waiting for upgrades before new capabilities
start to emerge. At best, the system will be upgraded with the designer’s imagination on how the
system will work (Hamada & Saito, 2018). Therefore, it is the human helping driverless trucks to
adapt, while understanding how the technology works. Once it is understood, automation is opportunity
to improve how safely and efficiently trucks are driven. It could, however, just make haulage just as
high-risk as it is today, or worse (Department of Mines and Petroluem, 2014b).

Data Outputs and Insights

Data produced by a machine has typically outweighed the ability of humans to remain in-the-loop
(Wiener, 1989). Being out-of-loop is driven by automated system that combines labels, numbers and
colors that contain various levels of meaning (Endsley & Kiris, 1995). Supervisors of automated
systems need to adapt to new data languages often cloaked as machine insights (Sarter et al., 1997).
People follow recommendations given by a machine, with little insight into how it arrived at its
conclusion (Hurley & Adebayo, 2016). Automated systems are marginally transparent, given that their
algorithms are considered Intellectual Property of the designer (World Intellectual Property
Organization, 2019). Therefore, automated systems operate independently from their users, limiting the

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 9

decision-making process to enable user’s ability to solve their own problems (Brantingham et al.,
2018). While the automated systems gather performance data, the information is fed back to the vendor
for optimisation (Hurley & Adebayo, 2016). Optimising the system without educating the end-user to
truly understand what is happening, limits the overall human-centered improvement cycle (Giacomin,
2015). Transferring driving tasks to a machine, redistributes all the tactical information previously
communicated by truck drivers (Caterpillar, 2013). Where a simple discussion could be held with a
truck driver, now needs to be carefully extracted from filtered information or collected by observing
truck movements (Caterpillar Global Mining, 2019).

Driverless technology appears to have skipped ahead of the research theories that try to explain data
ontology. The WA Mining Industry automated trucks whose agency they previously understood and
actively controlled (Department of Mines and Petroluem, 2015b). However, as haulage systems were
engineered, the interaction and dependencies on others change, yielding a more complex by-product
(Department of Mines and Petroluem, 2015a). Vehicle interactions evolved from complex situations
(Department of Mines and Petroluem, 2015c); requiring equipment operators to use data to navigate
and foresee situations. Understanding system data is unique to that person and cannot be reconstructed
by any other person (Sieck et al., 2007). As more driving tasks are automated, the further people are
removed from the immediate process (Wessel et al., 2019). Therefore, data outputs can become more
mysterious than they were previously (Rankin et al., 2016). Although the developer designs the system,
it is the user who is responsible for working out what the data trying to explain (Endsley, 2016). The
problem is that interconnections form to quickly distribute data across the system, informing and
directing people on what automated equipment will do next (Christoffersen & Woods, 2002).

The lack of transparency in automation has not stopped designers from attempting to ‘augment’ human
work (Araujo, 2018). Augmentation is to enable people to be more creative and thoughtful by
computing data insights (Hebbar, 2017). Big data does all the heavy lifting, while a human simply
actions the recommendations from the machine (Hurley & Adebayo, 2016). However, a solution-driven
approach is argued to deskill the human, rather than increase their knowledge and understanding
(Bravo Orellana, 2015; Ito & Howe, 2016). As people are promoted to higher levels of supervisory
control, the less they learn about the operation (Sarter & Woods, 1995). Whether data informs people
on what to execute, or simply supervise a machine to perform a task, people must identify situations
that are beyond the machine’s data set (Skeem & Lowenkamp, 2016). This activity is often under-
specified and requires people to improvise, reintroducing them into non-designed situations (de Visser
et al., 2018). The data presented can appear more confusing than it did before, making it difficult to
return the system to a safe state (Stensson & Jansson, 2014). Although display functions may list rules
that decided an outcome, the literal representation during peak periods may be cognitively restrictive
(Cummings et al., 2016). Nonetheless, providing users with access to computations and partial
decision-making are more useful than solutions (Zittrain et al., 2018). Furthermore, if the aim of using
data to is to augment human work, then the human must be creative with that data. While the ontology
is biased (Bolukbasi et al., 2016), opaque (Buolamwini & Gebrum, 2018), solutionist (Ross & Swetlitz,
2018) and specialised (Skeem & Lowenkamp, 2016), data will continue to reinforce old habits and
underpin the optimisation problems that led them to automation.

Literalism and Parameterisation

Driverless haul trucks give the impression they are safer by projecting their haul route. A truck displays
its travel pathway via an in-cab display in manually operated equipment (Caterpillar Global Mining,
2019). This level of transparency in travel routes can also increase the level of trust people have for
automated systems (Botsman, 2017). Providing travel pathways increase the certainty around what the
machine will do next. However, despite this level of transparency, a manually operated machine with

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 10

an in-cab display collided with a redirected haul truck (Department of Mines and Petroluem, 2015c).
This incident questions whether people are even observing intended truck routes. The redirected
pathway, following a corresponding truck slide, highlights the operating boundaries of automation
(Ockerman & Pritchett, 2002). Automated systems work well under designed conditions, yet they
perform poorly when situations are beyond their design parameters (Billings, 2018). These situations
can be compounded by machine logic that is able to identify a hazard, however, is unable to provide a
safe way forward (Caterpillar, n.d.-c). Operational boundaries are attributed to a set of constrained
instructions, which are insensitive to the continuous shift in priorities and objectives (Vul et al., 2014).
While attempting to achieve an objective, brittleness upsets the process, with an inability to execute its
part of the process (Ockerman & Pritchett, 2002). Therefore, the technology simply ‘throws up its
hands’, effectively stopping the process or immediately handing back control (SlashGear, 2017).

The value of human supervisors is their ability to exercise unconstrained thinking (Lake et al., 2015). A
human is capable of drawing information from external sources to improvise during novel situations.
Leveraging from previous experiences allows humans to solve localised problems, which machines
may be incapable of solving (Reason, 1990). Despite the benefits of a truck performing “exactly as the
computer has programmed it do” (ADVI Hub, 2016), the downside is that they perform nothing else.
Systems that are predictable, are not overly adaptable (Inagaki, 2003). Therefore, it is the system-based
roles that cover the shortfall (Dekker, 2003). This trade-off raises an interesting conundrum of the
value of replacing human work. With constant supervision and intervention, the value proposition
diminishes (Noy et al., 2018). Despite the direct safety benefits of removing humans from a high-risk
task (Palmer, 2019), the consequences are only just coming to fruition (Department of Mines and
Petroluem, 2015c). After all, automation is ‘stupid’ (Domingos, 2015), exacerbating the reliance on
humans to provide the context to make informed decisions. However, this can be difficult, with
literalism restricting the reconfiguration of instructions (Billings, 2018). Difficulties arise when
attempting to redirect an automated system. Not only enabling the machine to understand new
instructions, yet to perform those instruction as expected (Woods & Hollnagel, 2006). Without them,
the operating parameters simply retain the status quo, hiding the limitations of the logic while
everything else adapts (Winfield & Jirotka, 2017).

Automated systems can appear adaptive when compared against the data set it was trained on (Prechelt,
2012). For example, parts of the programming data are held out for testing, so when machines are
tested against humans, the performance appears comparable (Walker, 2016). However, the
performance is not comparable, particularly when tested on non-training data (Buolamwini & Gebrum,
2018). This logic applies to driverless haul trucks and their ability to recognise objects (Caterpillar,
n.d.-c). LiDAR and Radar technology are capable of identify objects, yet they are unable to distinguish
between windrows and people (Teichman et al., 2011). Parameterisation problems become apparent
when Global Positioning Systems (GPS) on driverless trucks are ineffective, which has reverse obstacle
detections fail (McKinnon, 2019). Moreover, self-driving motor vehicles experiences a similar
phenomenon, where the oncoming vehicle was unable to identify a pedestrian in time to stop (National
Transportation Safety Board, 2018). Despite driverless technology coming along way, it is not there
yet. There are more advancements to be made, with various industries that need to be made aware of
machines functions and their technical capabilities (Payre et al., 2016).

Protectionism and Resilience

Engineering defence layers to protect a system from failure is widely accepted practice in risk
management (Summers, 2003). Defence in-depth is philosophy grounded by intelligence hardware that
assumes incidents unfold in a linear chain of events (Murphy, 2016). If this were the case, the more
connected a human-machine system became, the more redundancy would be required to counter

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 11

predictable domino-like reactions. The empirical basis for validating the success of controls are the
absence of significant events; if the system has not had an incident then the arrangements are safe
(ADVI Hub, 2016). Though this assertion can be misleading, Reason (1990) explains how automated
systems are not known for their response to isolated hazards. As an example, a virtual intersection that
is not demarcated in the physical mine laid dormant until a driverless vehicle needs to use it
(Department of Mines and Petroluem, 2015c). Retrospectively, the installation of signs and devices
could have assisted the human operated machine to identify potential interactions and avoid the
collision. However, the physical demarcation of every intersection simply adds more layers of
protection in already complex system, opening more disparities between the physical and virtual
environment (Caterpillar, n.d.-b).

To safeguard against incidents, engineers often design extensive levels of protection to create new
forms of failure (Caterpillar, 2013). Traditionally, the WA Mining Industry has prioritised layers of
protection over resilience, implementing theoretical walls that are incapable of bouncing back (Willey,
2014). The rigorous test structures and fail-safe systems implemented as a means of insulating
driverless technology from conventional incidents, seem to have created their own pathways that have
mystified the industry (Department of Mines and Petroluem, 2014b). Technology introduced to replace
human limitations (i.e. driver attention, concentration, fatigue) with layers of artificial intelligence (i.e.
LiDAR, radar, pattern recognition) have now become the industries greatest weakness (Department of
Mines and Petroluem, 2014a; 2014b; Teichman et al., 2011, 9-13 May). Perrow (1997) explains how
the fallacies of “defence in-depth” can obscure the view on how systems behave when they are
stretched and compressed. The result had left investigators puzzled how the driverless system became
so opaque to those who use them. The regulator reported a “lack of system knowledge and
understanding of how the autonomous equipment system works” (Department of Mines and Petroluem,
2014b, p. 1). What automation taught early adopters of automation, is that the more layers that are in
place the more domain experts are removed (Billings, 2018). When users are reintroduced back into the
control loop to solve system malfunctions, the processes can appear more peculiar than they did before,
making the recovery method process much more difficult (Pritchett et al., 2013).

Assisting people to cope with complexity is at the heart of resilience engineering (Dekker et al., 2008).
Technological innovation in the WA Mining Industry has resulted in dramatic improvements in
decreasing injury rates since driverless technology was introduced (Caterpillar, n.d.-a). Nevertheless, it
takes time for automation to magnify the inefficiencies in a process, even if the industries processes
were benchmark in both safety and productivity (Bellamy & Pravica, 2011). The role of the human is
radically re-engineered to remain the critical interface between sub-systems of complex whole,
particularly if they are dealing with multiple ‘expert’ systems with various objectives and limitations
(Fridman et al., 2018). How well a system withstands variations and disruptions outside of the design
envelope is an indication of how resilient it has become (Chandler, 2014). Human flexibility and
adaption are yet to be truly understood by cognitive scientists, with various skills sets to be engineered
into a machine (Lake et al., 2014). Machine learning may be able to beat the world’s best AlphaGo
player; however, it still cannot drive to the match (McFarland, 2017). For a system to be agile and
successful in this digital revolution, it must mature beyond machine literalism to be pivot and
manoeuvre around danger (Srinivasan & Mukherjee, 2018). Ito and Howe (2016) believe that
augmentation holds the key, fostering the relationship to create the foresight to anticipate risk and
navigate the complexity of ever-changing landscapes.

Manual and Automated Modes

Whether a haul truck is in manual or automated mode depends whether it has been programmed into
the machine. Moreover, if system engineers are yet to figure out how to automate the task, then trucks

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 12

must be operated in manual mode. This can also be said for communication losses, where network must
be maintained in order to control trucks automatically (McKinnon, 2019). A mode can be identified
through the lighting system for ancillary equipment operators and via in-cab display for system-based
roles (Caterpillar, 2013; Caterpillar Global Mining, 2019). While some mining operations mix manual
and automated haul trucks, others choose to separate the differently operations entirely (Department of
Mines and Petroluem, 2015a). This approach can alleviate the confusion behind determining whether
the truck is manual or automated mode (Sarter & Woods, 1995). More importantly, the different
functions and rules associated either mode (Glover, 2016). Endsley (2016) explains how the
consequences emerge when people are surprised by equipment functions, which can ultimately lead to
haul truck interactions. Alleviating the issue requires improving the dialogue on the overall objectives,
operating envelope, next movements and resolution logic (Salas et al., 2010). Feedback loops are
considered to be a starting point to merging the gap (Sklar & Sarter, 1999). Consequently, improving
feedback could the minimise short phases of intervention, observed as the frequent cause of people who
have lost track of machine assignments (Feldhütter et al., 2019).

Human factors research appears optimistic on the progress towards user comprehension of automated
modes and configurations. Norman (2013) argues that the idea is to put knowledge into the world.
While some academic papers promote the development of rich ‘mental models’ for automated systems,
Sarter (2008) contests that the theory is empirical flawed. Regardless of what product vendors train
their users on what to look for (Merritt et al., 2015), Sarter et al. (1997) claim that there will always be
mismatches in the way humans supervise machines. Automation surprises are argued to be a normal
by-product of a machine that undertakes work independently (de Visser et al., 2018). If a system
required ‘safety drivers’ for motor vehicles, for example, the productivity value would soon diminish.
However, when a driverless machine confronts a novel situation, it can become quite onerous when
attempting to draw from external resources (Endsley, 2016b). Identifying the correct mode is a
consequence of the system’s design, not the fact that automation has gone too far (Norman, 2013). A
mismatch in mode identification occurs when the machine’s interface does not visibly display the
mode, which requires users to remember a mode from hours earlier (Feldhütter et al., 2019). Casner et
al. (2016) emphasised that designers should allow for possible intervening situations that can distract
humans from remaining in touch with the machine’s mode of operation.

The transition between manual and driverless control has been identified as an unconventional risk for
the WA Mining Industry (Department of Mines and Petroluem, 2015a). Automation can generate
unanticipated changes in a haul trucks’ direction, leading to a loss of vehicle control. If the loss of
control had indeed involved a mode change, then it is likely that this situation was recorded somewhere
in the system. Björklund et al. (2006) explain how retrospectively, available data gives rise to
engineering confidence that transitions are observed by human users, with the view that higher
attention rates can avoid such occurrences. Nevertheless, researchers are left with a puzzling thought
when people, who do not communicate with machines, understand what mode a machine is in. Dekker
(2014b) suggests comparing the difference between the machine function and user’s intentions, the
disparity and similarities offers some indication of the persons’ awareness.

3.2 Research Question 1

How are the theoretical viewpoints of human-machine systems influencing the approach to haul
truck automation?

The theoretical viewpoints are underpinned by science and engineering. Both fields study systems by
‘reducing’ them to their most basic parts, analysing what is contained. Engineering attempts to replicate
the components by reverse engineering human tasks. Machines are then programmed with the patterns
that are recognised in basic human level performances. The technological advancements made
https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 13

available are then introduced into the system, reallocating activities to either human or machine. The
system is theoretically ‘reconstructed’ by the designer, specifying tasks to be undertaken by driverless
trucks (i.e. drive, load, tip) and residual work by humans (i.e. object clearance, surveys and
assignments). When the system is theoretically reconstructed and put back together, existing
relationships and connections are transformed to create new situations like the watercart incident (see
Department of Mines and Petroleum, 2015c).

3.3 Processes in Human-Machine Systems

Opacity and Transparency

Augmenting the relationship between driverless trucks and their supervisors depends on the
technology’s transparency. Opacity is the by-product of a highly protected technology that reduces
human capacity to comprehend its function (Billings, 2018). Therefore, the system provides minimal
insight into how the algorithm decides an outcome (Dressel & Farid, 2018). For example, driverless
trucks may perform U-turn at a loading source without notifying human supervisors on the reasons
why. Demystifying the opacity requires the illumination of the decision-making process (Winfield &
Jirotka, 2017). According to Wiener (1989, p. 244), pilots of automated aircraft frequently asked:
“What is [the machine] doing? Why is doing that? What is it going to do next?” Automated systems
can even be deliberately designed to limit their transparency. One of the reasons is to protect the
designer’s Intellectual Property (World Intellectual Property Organization, 2019), while another is to
avoid the technology from being overridden. However, the consequences leave humans unable to track
the machine’s mode of operation (Sarter et al., 1997). Furthermore, the unanticipated actions of the
machine can result in automation surprises (Woods & Sarter, 1998). Automation surprises are
anticipation of one action (turn left), yet the machine performs something different (turns right). These
surprises where previously alluded to by Norbert Wiener. His study of B-757 pilots found that 69% of
participants were surprised by the automated actions, while 35% were unsure of the technology’s
modes and features (Wiener, 1989). This phenomenon was replicated in the WA Mining Industry,
where an investigation found that people involved in a driverless truck incident had a “lack of system
knowledge” (Department of Mines and Petroluem, 2015c, p. 1). Despite the protection of a truck’s
decision-making process, it appears the trade-off is stifling the creativity and understanding of
driverless truck functionality.

The processes used to collaborate with machine can become increasingly vague to humans. Particularly
as the technology evolves and progressively replaces more human work. The more people are
promoted to a higher level of supervisory control, the more they are removed from the immediate
process (Stanton et al., 2001). Moreover, the greater number of trucks that are automated, the smaller
number of people available to understand suitable driving techniques. The intricate knowledge of a
truck’s gear range, turning circle, reverse capability and handling will be minimised. Therefore, the
transparency of the system’s capability will become increasingly important, explaining how the truck
performs routine tasks. Contrastingly, in order to compensate, humans learn by observing how the
truck behaves. Automation typically filters out direct information that explains the reasons for those
actions (Zittrain et al., 2018). Consequently, the user implements more test structures to verify
compliance to existing systems, simply adding more complexity and opacity to already multi-faceted
piece of technology (Department of Mines and Petroluem, 2014b). Despite comprehensive training
programs, Woods (1996b) explains how traditional training approaches may interfere with current
monitoring routines and learned interpretations of automation functionality. Providing transparent
feedback can be significant challenge, with interfaces required to provide vital pieces of information.
The balance is presenting information people need, without overloading them with information they
don’t need, or know how to interpret (Salas et al., 2010). Warning signs can become hidden among

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 14

complex web of information, the risk being no response at all (Dekker et al., 2008). The reverberations
of opacity are a false sense of security that processes are working as intended (Rasmussen & Vincente,
1989). However, if the transparency is there, human supervisors may be able to provide the
improvisations that they are designed to provide.

Tight and Loose Coupling

According to Perrow’s Interaction/ Coupling Chart, conventional mining techniques are loosely
coupled and highly complex in their interactions (Perrow, 1984). Prior to driverless technology,
haulage operations contained conventional buffers with flexible tendencies that the industry came to
understand (Department of Mines and Petroluem, 2015b). However, when driving responsibilities were
transferred to an automated system, haul truck connections with others had changed (Department of
Mines and Petroluem, 2015c). Where positive communication would be utilised to pass a haul truck,
now requires the truck to be virtually locked before passing (Hansen, 2020). Moreover, if a driverless
truck is assigned to tip at the crusher, the truck will remain stationary until it is cleared to tip, regardless
of time (ADVI Hub, 2016). Since the algorithm propagates across the entire system, every truck
performs the exact same activity. Such a highly connected system exacerbates the literalist thinking of
a machine (Dekker et al., 2012). Therefore, supervisors must think quickly to change functions and
instruct the automated system on what to do next (Miller & Parasuraman, 2007). Contrastingly, trucks
drivers who notice the crushers’ unavailability, simply ask the control room for another dump location
(BHP, 2018). The flexible tendencies of a human to adapt and ask questions, enables the haulage
system to become free flowing. Similarly, the situation occurs in losses of network communication
(McKinnon, 2019). In a manual system, truck drivers could operate if communications were lost.
However, for a driverless truck, the technology simply cannot operate unless communications are
maintained (Hamada & Saito, 2018). While automated systems are constrained by a narrow set of
objectives, the impacts of tighter coupling are experienced more rapidly (Jamasmie, 2019). Therefore,
the inefficiencies and failures have a greater impact and are much more difficult to isolate.

Automated systems are historically known for introducing characteristics that produce ‘normal
accidents’ (Perrow, 1984). An incident is considered normal when it involved normal people,
completing routine work, under normal circumstances (Wears et al., 2015). The focus is often at the
sharp end, arbitrarily reconstructing the sequence of events to evaluate human responses (Dekker,
2014a). The further the investigation moves back from the sharp end, the more coupled and connected
interactions become (Weber & Dekker, 2017). Therefore, systematic explanations are often replaced
with what was observed (Drury et al., 2012). This is where the notion of direct causes narrows our
thinking, the tight connections in a complex system are oversimplified (Department of Mines and
Petroluem, 2015c). As a consequence, the reductionist thinking leads to a broken component, while
other latent and tightly coupled aspects are underrepresented (Dekker, 2010). For example, a driverless
truck may slide out of lane, yet the loss of control could have been created by communications, traction
controls, speed zones, wet weather or road material (Department of Mines and Petroluem, 2015a).
Since the human response is to go after what did not work as intended, they immediately focus on
failure (Hollnagel et al., 2015). However, coupling is about focusing on the interactions themselves, not
the components themselves (Wears et al., 2015). In addition, all the components may have behaved
successfully. Therefore, safety lies in the interaction in tightly coupled systems, not the perfectly
engineered component (Hamada & Saito, 2018). Systems must be flexible, nimble and robust if they
are to navigate the complexities of the interactions they face (Cilliers & Presier, 2010).

Explaining the non-linearity of interactions does not prevent vendors attempting to provide solution-
driven products. Despite driverless capabilities being developed, the automated vehicle is unable to
effectively communicate with the crusher (Hitachi, 2015). Transferring control to a machine can

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 15

exemplify the inefficiencies that are contrary to the technology’s original intent. For example, without
information being shared between driverless trucks and the crusher, the reverberations of queue time at
the crusher can be enormous (Brundrett, 2014). The impact on people is that they are now being
required to intervene and reassigned the truck fleet. Although success is celebrated when technology is
componentised into a supple chain (Rio Tinto, 2018), automation eventually reaches its peak of
innovation (Panetta, 2019; Trudell et al., 2014). Eventually, technology becomes so standardised that
supervisors forget that systems’ defences can only protect against known causal pathways (Reason,
1990). Perrow (1984) points out that it only takes two or more components in a tightly couple system to
interact unexpectedly. As an example, it was unexpected that a driverless truck was unresponsive
towards a manual watercart, which was heading for its pre-defined pathway (Department of Mines and
Petroluem, 2015c). The non-linear reaction towards the watercart was under-specified relative to its
relationship, an oversight that caused a near fatal collision. And yet this problem would never have
occurred to the designer who has designed further collision and avoidance systems. As a result,
additional control systems can simply tighten the system’s coupling, while opening up more possible
interactions and pathways to failure.

Centralisation and Democratisation

Standardising residual human tasks is based on the predictive capacity of the designer. A capacity that
assumes centralising the most basic steps can guide supervisors to the safest outcome (Dekker, 2014b).
However, in a human-machine system, work instructions come with a caveat. A proviso that expects
people to follow written instructions, yet improvise when operational practices demand it (Dekker,
2003). Reason (1990) explains how the ‘Catch 22’ of supervising a machine cannot be escaped:
“Human supervisory control was not conceived with humans in mind. It was a by-product of the
microchip revolution.” (p. 2). As a consequence, the by-product is the result of designers unable to
predict and plan for every contingency (Caterpillar Global Mining, 2019). Despite this, Domingos
(2015) claims that his Master Algorithm will eventually equip machines with every contingency. Until
then, human intuition must inject smooth layers of local adaption, pulling information outside of
centralised sources to manage unanticipated situations (Pettersen & Schulman, 2016).

Spending countless hours training people in standardised methods is a common thread in safety. The
assumption is that standardising methods will build a cognitive repertoire to combat irregular situations
(Dekker et al., 2012). Moreover, designers will argue that their automated system has figured out it all
out, and there is no need for human intervention (Dietvorst et al., 2016). However, when the machine
malfunctions, supervisors must intervene in situations they may not truly understand (Tech Light,
2016). Reason (1990) made the point that automation denies machine supervisors the opportunity to
practice their post-automation skills, which ultimately leads to degeneration of domain expertise. When
human supervisors are eventually relied upon, they perform poorly (McKinnon, 2019). For example,
driverless trucks may not need human assistance for hours, then suddenly required to clear a reverse
object. In order to democratise their automated system, Toyota built their process from the bottom up.
The purpose was to increase their effectiveness and quality of workmanship (Trudell et al., 2014). A
company cannot “… simply depend on machines that only repeat the same task over and over again.”
argued Mitsuri Kawai, Toyota Executive Vice President (Mols & Vergunst, 2018, p. 122). Therefore,
automated systems may be efficient; but they are not overly skilful. Reverse engineering human
mastery in a machine will eventually become redundant (McCarthy et al., 2000). Thus, to compete with
low cost companies, industrialised nations are realising that their prosperity resides in user-centred
innovation (von Hippel, 2005). Improving a company’s supply chain may mean cultivating their inner-
Artisan, returning to the days of human craftmanship (Protzman et al., 2016).

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 16

Historical experiences do not account for truly novel events. A procedure detailing every design aspect
of the process does not always reflect the limitations of automated systems (Pritchett et al., 2013). For
example, the actions of machine supervisors labelled as “not adhering to…” or “failure to respond…”
may be an indication of the creativity required to get real work done under technological constraints
(Department of Mines and Petroluem, 2015c). In contrast to compliance-based approaches, perhaps the
use of procedures as recipes can democratise the system enough for the people to continuously
innovate (von Hippel, 2005). As a result, processes can then leverage the problem-solving aspect of
human intelligence, therein be more impactful than debating deviations from centralised procedures
and contrasting individual experiences (Lake et al., 2016).

Virtual and Real-World Distinctions

Representing the physical world through virtual maps may suggest to human supervisors that the
systems’ interface is a true. Supervisors may also believe that physical controls are in place simply
because the virtual representation displays it (Caterpillar, n.d.-b). Research surrounding the distinction
between physical and virtual worlds however, points to something different: an ideal world that is free
from localised constraints (Dahlstrom et al., 2009). For example, virtual displays can be a supreme
worldview how the system should look and function from an engineering perspective. Salas et al.
(2010, p. 10) argue that real-world problems are “far removed” and are replaced with simplistic
representation. It is essential that virtual representations co-evolve, seeking human input as they
attempt to solve frontline problems (de Visser et al., 2018). Local constraints consist of many different
parts, which can produce surprising and unpredictable situations for the user (Sarter et al., 1997). Thus,
when physical changes that are not retrospectively updated, the condition may not visible to the user to
warn them of an upcoming intersection (Department of Mines and Petroluem, 2015c).

The regulator governing mobile autonomous mining systems in Western Australia (WA) appears fairly
pessimistic about the representation of physical mines. The Department of Mines and Petroluem
(2015a) highlight a number of hazards associated with integrating driverless machines into an existing
environment, recommending a phased approach to the introduction of advanced technology. The
segregation of manually and automated haul trucks is designed to manage the risk of virtually and
physically controlled interfaces. Although the designer may have developed tools to redesign the
virtual system to meet operational needs, technology cannot remove the problems that technology
creates (Baxter et al., 2012). The challenge of pre-programming a machine is that operational problems
just keep moving, pushing the innovation curve outside of the automated systems’ pipeline (Trudell et
al., 2014). Analysing what a process contains does not explain what it will do, which makes updating
virtual displays a never-ending iteration (Woods, 2016). Moreover, representative samples of the
physical world can differ from human perception, which are constantly re-framed for meaning and
insight when displays are not in real-time (Rankin et al., 2016). Given the complexity of representing
the physical world, the on-board computational requirements for automated systems are extensive.
Therefore, there is a need for more computer power than what can physically fit on a machine, given
the amount of data processing required to operate LiDAR, image recognition and radar technology
(Goel, 2016).

Processing data gathered from vehicle sensors is critical to keeping visual representations real. Road
network surveys allow a virtual road map to be created (Teichman et al., 2011). The location of each
‘connected’ vehicle can then be tracked against the virtual model to determine the vehicle’s speed and
direction (Hamada & Saito, 2018). Automated and manually operated vehicles can then identify the
proximity of other vehicles, providing both agents with the means to reduce potential interactions
(BHP, 2017). System supervisors are also given the capability of implementing virtual speed, traction
zones and clearing obstacles (Caterpillar, 2013). Virtual zones allow users to make the connection

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 17

between surveys and surfaces in line with the physical environment. Supervisors can also control the
speed of the vehicle in the event that a machine is unaware of changing weather conditions
(Department of Mines and Petroluem, 2015a). However, as previously discussed, the machine will do
exactly what it has been programmed to do. Consequently, if a virtual zone has been surveyed beyond
the physical boundary, a driverless machine will still attempt to drive to those parameters (Department
of Mines and Petroluem, 2014b). Moreover, if a truck loses communication, the virtual mine model can
only identify its last known location. In the event of an interaction, the truck is now considered an
object and has the potential to cause a collision (McKinnon, 2019).

Active and Passive Workload

Transferring control to a machine may appear like a logical step to reduce human workload. Perform
lots of analysis, work out the most effective method and then engineer those actions into a machine
(Lake et al., 2016). Although the assumption here, however, is that the underlying conditions that make
this method possible will remain unchanged. Eventually, an automated system will face situations
beyond its training set (Buolamwini & Gebru, 2018). Ferris et al. (2010) explain how the workload of
supervising machines are short intensive moments, backed up by long periods of inactivity. This
workload phenomenon was uniquely observed by Perrow (1984) to cause workload ‘bunching’.
Bunching the demands for human input is an error inducing mode of operation according to Reason
(1990). Moreover, humans can be faced with an influx of requests from a machine that may not even be
executing a better job (Endsley, 2017). Attempts to evenly spread human workload is often confronted
with more engineering (Dekker, 2004). Product vendors will claim that the user will always be in
control (Rousseau, 2015). However, a quick transfer of responsibility can result in negative outcomes
when humans are not equipped to take over control (National Transportation Safety Board, 2017).

Automating human techniques have been long argued as a performance optimiser than a workload
minimiser (Prewett et al., 2010). Nonetheless, efforts are still being made to reduce human input often
cloaked as ‘augmentation’ (Dressel & Farid, 2018). For example, a machine may be assigned to
analyse data and offer solutions, however users are not privy to inputted data and how it arrived at a
conclusion (Dressel & Farid, 2018). The inaccuracies of data prediction highlighted by Brantingham et
al. (2018) and the clumsiness of automation noted by Lee and See (2004), undermines a supervisors’
trust. Constantly verifying a machines’ decision-making process is a highly cognitive task, meaning
that humans will avert using algorithms altogether (Dietvorst et al., 2016). The workload of machine
supervisors is argued to be a normal by-product of an automated system that proceeds without user
input (Miller & Parasuraman, 2007). Contemporary research in cockpit automation found a misleading
conclusion on workload, noting that automation is not ‘autonomous’ and cannot always be left to its
own devices (Edwards, 1977). Despite fewer physical activities being performed, the cognitive
demands of monitoring a computer system actually increases (Wickens, 2008). Moreover, it is less
likely that the intervention methods needed to recover a machine are not memorised, nor would they
unfold as the training proceeds them (Engle, 2016). The main driver for automation is not reducing
workload per se, rather making a process safer and more productive (Yeomans, 2014). Therefore, the
more reliable automated machines become, the higher the expectation to improve their performance
will become.

Cognitive overload has contributed to many incidents in Aviation. Flight deck incidents have occurred
in systems where human workload was thought to have been reduced (Wickens et al., 2016). For
instance, an automated system failure led to pilots’ performing a manual calculation for the aircraft’s
landing. At the same time, the pilots were unaware of the parallel problems of a single engine
malfunction. Although the pilots eventually responded, the wrong engine (the only functioning engine)
was subsequently shutdown (Salas et al., 2010). Prior to this event, the Aviation Industry would have

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 18

celebrated the reallocation of workload to a machine. Allocating work to a machine is argued to relieve
humans to focus on more important tasks (de Winter & Dodou, 2011). Yet, human users still find
themselves monitoring a machine’s activities for non-designed situations (Victor et al., 2018). The
residual is a bi-directional bridge between physical and cognitive tasks, manoeuvring among
monitoring and taking control over control in order to remain in touch with local constraints (Casner et
al., 2016).

3.4 Research Question 2

What processes are designed to support automation to equip human supervisors to improvise
during non-designed situations?

The processes of automation are residual tasks that the designer is yet to figure out how to automate.
The processes work well when the system is performing as intended. However, when faced with novel
situations, the processes are unable to be adapted beyond their design parameters. Since the designer is
unable to imagine and prepare for every contingency, human supervisors must use their unconstrained
thinking to draw from external information and previous experiences. Therefore, the processes work
well in designed situations, yet lack the relevance and adaptability when situations do not unfold along
pre-determined lines.

3.5 Human Factors

Mode Awareness

Mode awareness is recognising a machine’s state and understanding its operational parameters (Funk et
al., 2009). Driverless haul trucks operate in three different modes: autonomous (solid blue);
autonomous-ready (flashing blue) or manual (green) (Caterpillar, 2013). Mobile equipment operators,
maintainers and system technicians must understand the functions of each mode, particularly when
mode changing a truck. Maintainers and system technicians are required to enter the truck’s footprint to
manually recover, refuel and inspect the machine (Department of Mines and Petroluem, 2015a).
Therefore, the truck is required to be switched to manual mode for the duration of the task. A system
interface located inside the light vehicle allows technicians to perform mode changes locally (Today
Tonight, 2018). Alternatively, Mine Control is contacted via two-way radio to switch the truck’s state
to manual mode (Glover, 2016).

Driverless haul trucks can operate in the mine in manual or autonomous mode. Manually operated
equipment must identify the mode of operation and satisfy the attentional demands. Sarter and Woods
(1995) claim that when designers increase automated mode functions without the support of human
cognitive requests, mistakes in mode identification is often the consequence. Errors in identifying
operating modes have been a factor in human-machine systems for decades (Monk, 1986). The
introduction of driverless haul trucks into a mining operation has the potential to replicate similar
mode-related incidents (Sarter, 2008). Confusion around what mode a machine is in is at the heart of
automation surprises, where a user instructs the system to do one thing, yet the mode allows it to
perform something different (Sarter et al., 1997; Wickens et al., 2016). Studies into mode errors in
Aviation have found that minimal system feedback, complex functions and mental models reduce mode
awareness of pilots (Björklund et al., 2006; Sarter & Woods, 1995). In addition, the testing of partially
automated vehicles is finding similar mode awareness problems in safety drivers, which identified a
lack of mode awareness being driven by monitoring inattention (Feldhütter et al., 2019).

The importance of mode monitoring of driverless trucks is to anticipate the actions of the machine.
Misconceptions can arise in a persons’ mental model of automated systems, which underpins the

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 19

expectation of what the system will do next (Salas et al., 2010). Mental models that are vague and
incomplete, invite opportunities for automated systems to engage in functions not assigned by users
(Rankin et al., 2016). Equipment operators are able to observe a haul truck’s assignment; however, they
cannot see the details of that assignment, performance restrictions or decision-makings. Instead, they
must rely on their mental model of the driverless truck’s function to manage the underlying process
(Hansen, 2020; Today Tonight, 2018). For example, a technician will be unsuccessful in attempting to
activate an emergency stop a truck manually controlled. Unlike automated motor vehicles, technicians
are not expected to immediately regain control of a truck (Kyriakidis et al., 2017). As a result,
driverless trucks that are unable to operate automatically come to a controlled stop and are driven
manually to a safe location for observation.

Responding to Information and Warnings

Supervisors of driverless technology must be capable of responding to information and warnings.

Information and warnings in driverless systems include obstacle detections, health events, proximity
detections and truck performances (CAT, 2020; Glover, 2016). Therefore, observing and acting upon
this information is critical to supervising automated systems. The modality of the information is
presented in various forms, including visual and auditory cues (Caterpillar Global Mining, 2019).
Investigations may find that supervisors of automated systems failed to respond to system warning. A
critical point in time when someone should have intervened (Department of Mines and Petroluem,
2015c). However, the information that was available, is not necessarily the information that was
observed (Dekker, 2014). For information to be observed, cognitive is work required to determine what
the system is trying to tell them (Woods, 2018). Woods and Hollnagel (2006) explain that observability
not only depends on visual displays, but on personal interests, workload, objects and attentional
demands.

Humans are not passive receivers of information; they are actively acquiring, sensemaking and acting
upon data. The basic ideology of information processing is surveying the surrounding environment and
comparing it to stored memory (Dekker, 2019). For the processing of that memory, Engle (2016)
considers Baddeley and Hitch’s (1974) working memory system as a temporary storage of information
that regulates attentional demands. When determining the relevance of that information, the process of
sensemaking fills the gaps between in what was anticipated (remembered) and what was observed
(stimulus) (Rankin et al., 2016). When a sudden mismatch occurs between the two, automation
surprises start to emerge (De Boer & Dekker, 2017). Information processing has historically been
modelled on computer functionality (Eysenck, 1993). Visual information was theorised to be a visual
scratchpad that is situated in a working memory. For example, Parasuraman (2000) proposed that
information was acquired, analysed, selected and responded to, through these four broad functions of
human processing. The functions could then also be used as a basis for automation (de Winter &
Dodou, 2011). This notion, however, has been argued as an arbitrary view on information sharing
among human-machine systems (Dekker, 2019). Researchers also argue whether input-output devices
should resemble human properties, as computer metaphors are artefacts that represent an over-
simplification of human thought (Stensson & Jansson, 2014). Processing information is not the only
problem, there are other collaborative issues such as transparency (Winfield & Jirotka, 2017),
explainability (Gunning, 2016), feedback (Sklar & Sarter, 1999) and literalism (Billings, 2018).

Computers are rarely transparent in what they are doing and how they got there (Skeem &
Lowenkamp, 2016). Technology often withholds the data sets that were used to decide an outcome.
This is a normal by-product of automated systems. When working with a strong and silent character,
the cognitive demands of interpreting its outputs are high (Christoffersen & Woods, 2002). The
purpose of data, however, is not just providing information per se, its assisting the supervisor to

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 20

understand what the machine is performing (Miller & Parasuraman, 2007). The critical test is when the
device helps humans notice more than what they were specifically looking for or expecting (Sarter and
Woods, 1997). The failure of this test is restricted to humans: not identifying information, observing
information correctly, forgetting data and negatively reacting (Dekker, 2014a). However, it is a much
more complex relationship between human and machine, not the sole processing capability of the
human to observe, analyse and respond to information (Woods & Hollnagel, 2006). If humans are
going to fulfil their role as machine supervisors, information needs to flow freely between human and
machine. The impact of responding to information on supervisory roles are significant, given that the
position direct trucks based on the system’s information (Caterpillar Global Mining, 2019).
Consequently, the available information has become an instrument to inform supervisors on what
driverless trucks are likely to do next.

Craftsmanship and Skill Degeneration

While machines are replacing humans in repetitive tasks, a level of Artisan craftsmanship must still be
retained (The Wheel Network, 2016). Domain expertise comes to fruition when a machine is unable to
resolve a non-designed situation (Endsley, 2018). While automated systems are not known for
improvising, the process they are repeating must eventually be improved upon (Trudell et al., 2014). As
a machine becomes more reliable, supervisors are denied the opportunity to practice their marginalised
existence (Berdicchia & Masino, 2018). The degeneration of skills forms a vicious cycle, where the
domain expert begins realising their own incompetence and dependency on machines (Bravo Orellana,
2015). Even though manual skills are mastered through practical application, recalling those craft-like
skills in an emergency are reduced (Li et al., 2014). Particular cases in automated driving point towards
an over-reliance on automation (Körber et al., 2018). Salas et al. (2010) noted that pilots became
heavily dependent on FMS-generated displays, which were reducing their ability to identify the
proximity of travel way points. More immediate information is supposedly available in conventional
methods such as flight charts. However, there is no real purpose of introducing advantageous
technology if the value of the product is not being realised.

Taking advantage of automation means fully understanding the tool humans are using. The uptake is an
indication of the trust people have in the machine’s ability to operate independently (Hoff & Bashir,
2015). Although Lee and See (2004) observed a high level of trust, the consequence was a much higher
dependency on automation. In contrast, a heavily manually operated system was a symbol of distrust,
resulting in lower levels of utilisation (Payre et al., 2016). When users manually control a system to
“help the robot through some situations…” (MIT Sloan CIO Symposium Videos, 2017), the local
adaptions can be confusing when solving beyond the control loop (Dekker, 2003). What procedure to
apply and when is the talent, especially when the recovery mission is novel, complex and the procedure
is arbitrary (Goteman & Dekker, 2007). Users discovering their own competence in the application of a
procedure can be misled, confronted by overlaps in the physical and virtual world that obscures the
‘truth’ (Reason, 1990). Furthermore, reflexivity is underpinned by the limitations of explaining failure
and how their bias impacts on relevance (Holroyd, 2015). The cognitive skills that are vital to solving
frontline problems are now on the peripheral, only ‘flicking the switch’ when needed.

The main reasons why humans are retained in automated processes is to help the machine through
‘blind spots’ (Noy et al., 2018). Aiding the machine meant that humans must also develop an adequate
‘mental model’ of how the system works (Strand et al., 2018). Product designers cannot imagine every
scenario that is likely to be encountered, even if machine learning can help robots learn various
scenarios from big data (Fridman et al., 2018). Therefore, users are often left to work out what the
machine is capable of and what it is not (Lynas & Horberry, 2011). Suddenly re-introducing humans
back into the control loop can leave them feeling disorientated (SlashGear, 2017). A quick transfer of

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 21

control in aviation is considered by Endsley (2016) to be risky, as pilots are not necessarily aware of
the situation that is arising. Recent evidence suggests that driverless processes are becoming so novel
and complex, that humans are performing negatively (McKinnon, 2019). Reinforcing supervisors in
residual recovery methods to combat non-designed situations may not even be relevant (Payre et al.,
2016). Task simulation can mirror the process through virtual reality, however there is no guarantee
that the situation will proceed in such a manner (Frimpong et al., 2003). Perhaps, it is not through big
data that machines will learn how to perform human work, rather through the coaching and mentoring
from the finest experts in the domain.

Intervention and Omission

People will always consider their ‘tinkering’ as a master stroke. Whether a supervisor is installing a
speed zone, managing the fleet’s saturation or pursing more tonnes for the day. Intervention is an
extension of demonstrating that they know more about the situation than the machine. Conversely,
designers view human intervention as an unnecessary step in the process (Caterpillar, n.d.-a). This is
due to the fact that functions are already allocated on human and machine strengths (de Winter &
Dodou, 2011). However, Dekker and Woods (2002a) rendered the MABA-MABA (Men-Are-Better-
At/ Machines-Are-Better-At) approach irrelevant for human-machine systems. This rationale is that
human-machine capabilities co-evolve over time. Not only do humans continuously learn how
driverless trucks perform, the technology itself is subjected to software upgrades (Today Tonight,
2018). Since the both capabilities are continuously evolving, the evolution could explain the types of
acts and omissions of observed on driverless mine sites (Department of Mines and Petroluem, 2014b).
For example, a software upgrade may no longer require supervisors to upload a survey, however the
automatic upload may not be suitable for use. Therefore, the human needs to intervene in order receive
accurate information. This type of localised intervention, however, is often seen as non-routine and
contradictory to standardised methods (Dekker et al., 2008).

Designers retain people in automated systems to monitor truck performances. A paradox emerges when
deciding whether to intervene in the situation or not (Dekker, 2003). When pre-empting failure,
driverless truck supervisors have the option to step-in and control the situation or allow the machine to
manage itself. For example, emergency stop devices can bring the fleet to a controlled stop, yet an
immediate stop can also generate its own set of risks (Department of Mines and Petroluem, 2015a). For
instance, driverless trucks can slide out of lane as they attempt to suddenly stop. Moreover, the
situation could be compounded if trucks were descending a ramp into an Active Mining Area (AMA).
Conversely, if human intervention avoids failure, then the act is seen as a mark of expertise (Reason,
1990). Then again, if the action is not in accordance with a procedure, it can be considered a non-
compliance towards the safety system (Dixon et al., 2007). When it comes to omissions, supervisors
can simply be following the procedure, despite foreseeing the potential dangers. This is where human
supervisors are held responsible for not intervening when they should have (National Transportation
Safety Board, 2018). However, people can be heavily influenced by increases in false alarms and
warnings (Wickens et al., 2009). This explains why safety drivers have turned off automated control
systems in the past (Coppola, 2018). Nevertheless, closing the performance gap of automated systems
is what intervention is striving to do, while omissions can be a sign that people are out-of-the-loop
(Endsley & Kiris, 1995).

On the inside, informal work processes are powered people connecting the dots. As previously
explained, procedures and RACI’s (Responsibilities, Accountabilities, Consulted and Informed) are no
more than the designer’s imagination of the system (Glover, 2016). Real work is performed along the
fringes through information systems and experimental invention (Protzman et al., 2016). Despite the
designers’ best intentions, there will always be instances where human supervision needs to help

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 22

machines through sight, touch and sound (The Wheel Network, 2016). Therefore, formal processes can
be scarcely inadequate to handle goal conflicts among the design and the application (Xu et al., 2007).
While standardised work collides with conflicting goal conflicts, the tension between the person
omitting just do their part, versus the intervention to ensure work quality, is heightened on the frontline.

Role Transformation

Driverless haul trucks have not only replaced truck drivers, automation creates residual roles and
transforms tasks on the peripheral (Caterpillar, 2013). Haul truck drivers now fulfil system support
roles, equipment maintainers or ancillary operators on transitioned mine sites (Palmer, 2019). Truck
drivers who were once active participants, now passively monitor driverless haul trucks through a
computer screen interface (Glover, 2016; Today Tonight, 2018). Monitoring automated systems is a
higher level of supervisory control, which expects humans to intervene intermittently during non-
designed situations (Banks & Stanton, 2016). People who may never have operated a computer before,
are now virtually adjusting lanes, installing speed zones and clearing obstacles (BHP, 2017, July 6).
Supervisory roles are not specially taught how to program a truck, they learn automated functions by
observing truck movements (Caterpillar Global Mining, 2019, Dec 17). The irony of learning functions
through observation is following the strict functional allocation, yet embody the improvised skills to
recover from system malfunctions (Baxter et al., 2012).

Truck drivers also have the option to become ancillary operators. Although the activities remain
manual, there are additional technological layers operators must learn (Caterpillar, 2013). Technology
demands that operators build a mental model of the system (Sarter & Woods, 1994), particularly when
operators are not involved in the programming. For example, grader operators may interact closely with
the truck to witness how the system responds. Learning by doing helps operators build their knowledge
base on automated systems. In addition, the introduction of mode lights requires ancillary operators to
understand the meaning of each mode (Today Tonight, 2018). There is also a screen located inside the
cab, which provides a predicted path for each driverless truck. Although the predictive capacity
increases transparency, it is another capability of observation and information processing (Parasuraman
et al., 2000). Traditionally, radio communication would be made in the event an ancillary machine
wanted to communicate with a truck (BHP, 2018). However, the control room is now contacted,
requiring trucks to be locked from moving before passing (Hansen, 2020).

The inclusion of system roles in automated systems is to aid robots through beyond design situations.
Endsley (2017a) points out, however, that humans are not overly skilful in responding to system
information. The reason is that supervisory roles are passive monitors of the system, suddenly handed
back control of a situation (Reason, 1990). Even with the unique ability to recall domain expertise,
supervisory roles are far removed from the immediate process (Miller & Parasuraman, 2007). In
addition, the information they receive is filtered by a computer interface (Fridman et al., 2018). For
example, intersections designed into the virtual mine model may not actually exist in the physical mine,
which can leave mine controllers none-the wiser (Department of Mines and Petroluem, 2015c). There
is a skill in locating information that is needed, when it is needed, while filtering through non-essential
information to determine what is happening (Endsley, 2016b). When re-introduced back into the
control loop, the recovery can become so complex and peculiar, that cognitive gaps in recovering the
system safely can emerge (Endsley & Kiris, 1995).

One apparent means of solving the problem is repetitively training people in system recovery and
diagnostics. Training humans to manage complex, opaque and tightly coupled systems can be difficult
(Billings, 2018). If it were possible to simulate and gameplay an extensive suite of emergency
situations, there is no guarantee that they would ever occur (Frimpong et al., 2003). Extensive use of
automated systems can lead to deskilling and over-dependence, reducing the cognitive and
https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 23

psychomotor skills that required for manual control (Parasuraman & Riley, 1997). Moreover, as
automated systems become more reliability, the less domain expertise is actually needed (Wickens et
al., 2016). Toyota expressed concerns over automation creating too many laymen and not enough
masters of the craft (Tech Light, 2016). By being so far removed, Bleicher (August 2017) explains how
the human craft reduces overtime. Despite technology endeavouring to augment human work, it can
also degenerate conventional skills and dependency on machines (Bravo Orellana, 2015). The
replacement of drivers undoubtedly transforms mine site work, with unconventional situations
confronting humans in their new formed roles (Department of Mines and Petroluem, 2014b)

Supervisory Control

Supervisory control was never conceived with humans in mind. Supervisors of automated systems
involve a residual set of tasks that engineers are yet to figure out how to automate (Caterpillar Global
Mining, 2019). More specifically, the role is in place to respond to non-designed situations to help
driverless trucks navigate around them (Hansen, 2020). For example, a driverless truck is capable of
identifying an object (Caterpillar, n.d.-c), however it is unable to clear or override the object
(Caterpillar Global Mining, 2019). The unrestrained ability of humans to solve problems underpins
their residual existence. Examining, monitoring and modifying processes that are otherwise executed
by automated systems (Miller & Parasuraman, 2007). While carrying out online problems, supervisors
are expected to monitor and tweak the system within the operating limits (Today Tonight, 2018). The
difficult component of this, is whether to intervene or not in signs of weakness (Dekker, 2003).
Supervisors can find themselves on a pathway to failure (Department of Mines and Petroluem, 2015c).
The catch is whether the intervention will be successful in avoiding the situation. It is also can be their
responsibility when they failed to intervene before an incident happened (National Transportation
Safety Board, 2018). In contrast, if their intervention is unsuccessful, the supervisor is often the one
who is accountable (McKinnon, 2019). Therefore, while ever automated systems are only responsible
for a narrow set of parameters, the role of the supervisory controller is expected to cover the latter.

Supervisors are not taught how driverless trucks are programmed; they learn by observing them. In
addition, supervisors are trained in how to work automation (i.e. press a button), not necessarily how it
works (i.e. algorithms, logic) (MIT Sloan CIO Symposium Videos, 2017). Therefore, if a driverless
truck performs something unintended, supervisors are not necessarily equipped with the knowledge of
the underlying logic (Hebbar, 2017). Although the role is specified, from a design perspective, the re-
introductions to control loops during novel situations are not (Endsley, 2016b). Non-designed
situations require human improvisations to perform outside the box (Reason, 1990). Enabling people to
work well under these circumstances, requires a collection of system knowledge, feedback loops (Sklar
& Sarter, 1999) and greater transparency (Zittrain et al., 2018). However, automation is not always
easy to work with, often described as an opponent rather than a team player (Christoffersen & Woods,
2002). Since the logic is fixated on achieving its goal, it will literally hold the ball until a human is
needed. Multiply this by thirty to fifty times, and this gives some indication of the monitoring demands
of a driverless fleet (Today Tonight, 2018). Automation is designed to operate independently, resulting
in the human monitoring needs falling to the wayside (Sarter et al., 2007). Consequently, the focus
becomes centred around the technology, other than user who is expected to assist the machine through
difficult situations.

Assisting automated systems has been described as being bunched (Billings, 2018). Workload that is
bunched is long periods of inactivity, followed by short intensive moments (Li et al., 2014). Human
workload can appear in these situations as the bottleneck, with the inability of supervisors to respond
and recover promptly (Prewett et al., 2010). Quite often, however, the machine has instantly
reintroduced them back into a novel situation. Suddenly, the supervisor is confronted with multiple

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 24

failures and is attempting to prioritise what should be done first (Miller & Parasuraman, 2007). Unlike
self-driving cars, where the safety driver is expected to take the wheel in any situation and at any speed
(Payre et al., 2016). Driverless trucks simply come to a stop wherever control is lost (International
Organization for Standardization, 2019). The difficulty, however, for supervisors of driverless trucks is
the recovery after a stoppage (Department of Mines and Petroluem, 2014b). For example, the task is
likely to be conducted remotely by Mine Control. In addition, field technicians and ancillary operators
become the eyes and ears to physically verify the situation. A combination of these roles enables the
driverless fleet to execute their daily tasks (Caterpillar Global Mining, 2019). Although certain tasks
are specified, situations emerge that require objects to be cleared (rock on road), surveys to be taken
(updating mine model) and instructions to be given (send truck away) at various times (Caterpillar,
2013). Therefore, there is a unique relationship that forms among humans and machines, and it is not
just those directly supervising the trucks either. The reverberations of supervisory control are as far
reaching as drilling, blasting, ancillary equipment, equipment maintenance and the control room
(Bellamy & Pravica, 2011).

3.6 Research Question 3

How does human adaptive behavior manage unanticipated machine performances and decide to
intervene or not during beyond design performances?

Humans adapt to unanticipated situations by drawing from external information and previous
experiences. Deciding whether to intervene or not is based on whether the supervisor believes that the
automated system will recover from the situation. External information such as radio calls, weather
forecasts and network systems provide external intelligence, while previous experiences of driverless
trucks navigating downpours, pit interaction and potential network losses indicate whether intervention
should occur. Interventions include speed restrictions, traction controls and setting changes. On the
surface, the adaptability of the human can appear unnecessarily tinkering to upset the automated
decision-making process. However, it is human who is held accountable if the outcome was negative
and the system supervisor was deemed to have the opportunity to intervene and avoid the outcome.

4. DISCUSSION
The literature has highlighted the fascination with creating and designing new products. Especially
when those products have the potential to advance the human race and provide a platform for
improving the way humans live their lives. However, the immediate approach to designing a new
product is to reduce the system into its most basic parts, separating the system into theoretical
components to determine how things work (Dekker, 2010). The problem with a reductionist approach
to understanding a system, is that a system is defined by what it does, not what it has. Designers are
instantly on the back foot, engineering a vehicle to travel from A to B with little knowledge on how the
mind made it possible (Victor et al., 2018). Moreover, the various paces of individual technologies
have limited the full deployment capability of some AI products. For example, a vehicle may be
capable of detecting an object, however it is yet to classify those objects for relevance (Held et al.,
2012). Such limitations in the real world has already led to car manufactures turning off automated
functionalities to accrue more travel time (Wakabayashi, 2018). Although researchers are attempting to
design technology that correctly classify objects in a vehicle’s travel path, the technology has a long
lead time for being deployed into a real-world environment. Moreover, testing similar technology in the
public sector has already highlighted the biases that exist in current engineering practices (Brantingham
et al., 2018; Buolamwini & Gebrum, 2018). The impact on driving could see vehicles classifying
objects incorrectly and applying the wrong functionality. For instance, the classification of a person for
a tree would ignore the fact that the person may cross the road. Furthermore, attempts to navigate a
road networks’ signs and signals with implied road rules is a significant task for a machine, given that
https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 25

it may not have been confronted with those variables through previous interactions (i.e. green light and
an emergency vehicle approaching) (Endsley, 2018).

The WA Mining Industry appears to be currently avoiding the complexities of object classification.
Driverless haul trucks do not attempt to distinguish between objects, rather stopping when the object
meets a size criterion (Caterpillar, 2013). This is a similar function to what Adaptiv claims to have been
turned off by Uber (Coppola, 2018). Since the technology struggles to distinguish between objects, the
vehicle would be constantly reacting to adversarial conditions on the side of the road (Eykholt et al.,
2018). By turning off the object recognition function, the vehicle could then travel uninterruptedly and
seek guidance from the supervisor only when required (National Transportation Safety Board, 2018).
The circumstances are, however, marginally different, with haul trucks unlikely to be carrying
passengers and therefore lowering the likelihood. Mining companies also have a team of well-trained
professionals who are taught how the processes support the technology (ADVI Hub, 2016). Although,
those processes are usually a set of residual tasks that were unable to be engineered into a machine.
Standardised processes are only as effective as the designers’ imagination, leaving the non-designed
situations up to human intuition (Noy et al., 2018). Sharing the control between human and machine
appears more realistic in the short term, becoming more transparent in the decision-making process to
allow humans to navigate the vehicles through complex situations. Despite this, Intellectual Property
and data protection concerns are stifling the pursuit of shared management (World Intellectual Property
Organization, 2019), the algorithms are at the heart of any business product (Mitchell, 2018). On the
other hand, for the technology to become truly ‘self-managed’, researchers and engineers must figure
out how the technology can be more adaptive like a human. Until then, designers will have to do more
to make shared technology more user-centered, allowing people to be more supportive in aiding
driverless systems through complex situations (Fridman, 2018).

The literature has highlighted, however, that technology has not always been developed with humans in
mind. The focus has always been to replicate and replace human labour, not partially succeed and allow
humans to take control of the system. Through the deployment of automation in industry though,
researchers have revealed that there is more to making a system work than allocating 25ecognize25ed
functions to various roles (Strand et al., 2014). Disruptions may arise that requires the function owner
to think outside the box. For example, if a weather system moves in, driverless machines and
automated aircrafts are currently ill-equipped to 25ecognize the changing conditions (Jamasmie, 2019).
The modes of communication between agents are not simple enough to explain that a weather system is
approaching either, requiring the automated system to change speed or direction. Obviously, if the
human supervisor was to argue ‘that the task was not their job’, the machine would likely put
passengers at-risk by functioning as if the weather conditions were not present. The interface between
human and machine is where this research investigation begins, not where it ends. Driverless
technology has changed the connections that manual driving had originally formed. There have been
numerous incidents on WA mine sites since the driverless technology was introduced, leaving
researchers wondering why. The lack of knowledge in this field provides focus and reasoning,
illustrating what research is yet to be fully understood and how objective findings can be drawn.

5. KNOWLEDGE GAPS
The reasons for incidents involving driverless haul trucks across the WA Mining Industry remains
relatively unknown. Although individual investigations may point out errors from either human or
machine, research is yet to explain the systemic influences of engineering a haulage system. For
example, there is more to a truck–on–truck collision than an inability of humans to respond quick
enough to a down pour of rain (Jamasmie, 2019). Certainly, having a process around the situation may
have coordinated the response to reduce truck speed. However, this observation is after the fact, neatly

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 26

joining the dots between the driverless machines’ limitations and the expectation of human supervisors
to manage the rest. It is often assumed that human supervisors will apply a smooth layer of local
adaption to fill in the shortfalls of automation; becoming the ‘eyes and ears of the operation’. Whether
a human should adapt a localised practice is not always clear, facing various situations that rarely
unfold in a predictable manner. Even though the assumption is that driverless machines are a like-for-
like replacement for truck drivers, this view could not be further from the truth. Not only do mining
companies transfer agency to the vendor when they automate the fleet, they appear to be left in the dark
on the decision-making process of their haul trucks (Mitchell, 2018).

This is where the gap becomes apparent. What was once a haul truck system that was under local
control is now transferred to a vendor’s central algorithm. The interactions change and require other
variables to adapt to the new relationships that are co-evolving on the mine. For instance, a truck no
longer makes a call over the radio to pass a working ancillary machine. Instead, a screen interface is
used to provide the intended route and alerts operators if they are too close. The different modes of
communication between mining equipment on a haul road demonstrates one element of the adaption’s
humans are making. The full capability of a driverless machine is not necessarily explained to the user
either, learning the strengths and weaknesses by observing its functionality over time. Therefore, a
driverless machine’s full capability is rarely understood upfront, leaning on local users to press the
buttons along the fringes to ‘feel out’ the machine’s parameters (i.e. what can this truck actually do?).
Although there are processes designed to support the system’s application, the processes are based on
how to work the system (e.g. press the button), not how the system actually works (e.g. how does it
function?). As the decision-making is not programmed by the user, the system has performed some
surprising functions. Those functions are not necessarily aligned to the users’ objectives either, driven
by the designers’ imagination and ability to reverse engineer best practices in mining. The impact of
the outcomes arising from automating machinery sketches a landscape where unique incidents start to
unfold, for which safety research must help the WA Mining Industry to understand.

With a backdrop of the arrangements that enable humans to be technically substituted for a machine,
the emergence of uncontrolled situations gives rise to the potential negative consequences. There is no
research exploring why truck slides out of lane and the potential those situations can cause. Perhaps the
sensitivity around new technology and the competitive advantage of being first, hinders the WA
Mining Industry’s ability to share lessons that are being learnt. Moreover, if the new risk profile of
driverless haul trucks is unknown, the risks can never be controlled. For instance, the LiDAR and
Radar systems on trucks are not capable of predicting slippery road conditions, what other mitigating
controls must be put in place? The explanation of the sequence of events and the contributing factors
that led to the incident are paramount when improving the safety system. Research must go beyond the
investigation findings that evaluates the trucks actions against its capability, which often reinforces the
common statement that the machine ‘did exactly what it was supposed to do’. If this approach to
understanding incidents was to continue, the WA Mining Industry’s knowledge will be forever
constrained by the world imagined by the product designer. The limitations of the technology and local
user adaptions that are taking place are forming new methods of work. The industry’s assumption is
that truck drivers are being replaced, and that technology removes the safety risks associated with haul
trucks. However, the entire process is far from being substituted, leaving a set of residual processes that
were technically difficult to automate. The interaction with driverless machines in operation still poses
risks to those who remain. Moreover, the uncontrolled nature of reversing a driverless vehicle over a
waste dump brings to mind other situations in which those circumstances could arise. Besides reporting
the cause as the humans’ inability to adapt, research must offer more constructive explanations to
manage risk, enabling the industry to develop effective systems of work when deploying driverless
haul trucks.

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 27

6. CONCLUSIONS
Exploring the experiences of other benchmark industries in their application of computerised control
systems is fruitful. The context in which negative events occur is important to examine, given the
transferrable similarities in the way the systems are designed and how professionals are using them.
Although there are many studies that consider the consequences of automation in various high-risk
industries, research is yet to comprehensively analyse what impact artificial intelligent machines are
having on WA mine sites. Furthermore, in light of recent events (McKinnon, 2019), understanding why
incidents involving driverless haul trucks are occurring in particular instances (Department of Mines
and Petroluem, 2014b). Thus, understanding the interactions between human and machine will explain
how the relationship is evolving in WA. Coinciding with theoretical models of the human-machine
relationship (Hancock et al., 2013), an examination of the contributing factors leading to incidents are
needed. This research endeavors to extend this knowledge through real-world examples, demonstrating
the causal pathways that have generated on a mine site.

The knowledge expressed throughout this study can inform the design of driverless technology, support
the formation of work processes and accommodate the local adaptions of human users. Previous studies
indicate that a human-centered design is central to positive performances in both safety and
productivity (de Visser et al., 2018). Researching the context behind a range of incidents involving
driverless vehicles has greater implications for the WA Mining Industry. The study highlights a range
of systematic trends that are not present in any one investigation. Furthermore, the analysis provides an
in-depth understanding of the phenomenon, which are often omitted and filtered when publishing the
investigation findings publicly. The underlying hypothesis of this research is that incidents involving
driverless vehicles are being shaped by WA Mining Industry’s assumptions, which has inflated the
expectation that the technology is a like-for-like replacement for haul truck drivers (Ernst and Young,
2019). However, as this study will explain, the technology is far from human-like. Despite its recent
advancements, local domain expertise continues to sooth the novelties along the fringes, while the
boundaries of its capability are continuously learn.

REFERENCES
ADVI Hub. (2016, June 27). Applying Automation in the Mining Sector to Environments in the Public Interface
[Video].YouTube. https://www.youtube.com/watch?v=dDAQoH1nlNU
Araujo, C. (2018). IBM bets big on the ‘incumbent disruptor’. CIO.
https://www.cio.com/article/3263695/digital-transformation/ibm-bets-big-on-the-incumbent-disruptor.html
Banks, V. A., & Stanton, N. A. (2016). Driver-centred vehicle automation: using network analysis for agent-
based modelling of the driver in highly automated driving systems. Ergonomics, 59(11), 1442-1452.
https://doi.org/10.1080/00140139.2016.1146344
Baxter, G., Rooksby, J., Wang, Y., & Khajeh-Hosseini, A. (2012, August 29-31). The ironies of automation ...
still going strong at 30? [Paper presentation]. 30th European Conference on Cognitive Ergonomics, Britain.
http://doi.org/10.1145/2448136.2448149
Bellamy, D., & Pravica, L. (2011). Assessing the impact of driverless haul trucks in Australian surface mining.
Resources Policy, 36(2), 149-158. https://doi.org/10.1016/j.resourpol.2010.09.002
Berdicchia, D., & Masino, G. (2018). Leading by leaving: Exploring the relationship between supervisory
control, job crafting, self-competence and performance. Journal of Management & Organization, 1-19.
http://doi.org/10.1017/jmo.2018.67
BHP. (2017, July 6). Jimblebar autonomous trucks [Video]. YouTube.
https://www.youtube.com/watch?v=NUfcPfh1Rcg

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 28

BHP. (2018, August 28). Ever wondered what it's like to drive one of the biggest trucks in the world? [Video].
YouTube. from https://www.youtube.com/watch?v=Ad1ZA6mrWgE
Billings, C. E. (2018). Aviation Automation: The Search for a Human-Centered Approach. CRC Press. http://
doi.org/10.1201/9781315137995
Björklund, C. M., Alfredson, J., & Dekker, S. W. A. (2006). Mode Monitoring and Call-Outs: An Eye-Tracking
Study of Two-Crew Automated Flight Deck Operations. The Internationa Journal of Aviation Psychology,
16(3), 257-269. http://doi.org/10.1207/s15327108ijap1603_2
Bleicher, A. (2017). Demystifying the Black Box That Is AI. Scientific American.
https://www.scientificamerican.com/article/demystifying-the-black-box-that-is-ai/
Bolukbasi, T., Chang, K., Zou, J., Saligrama, V., & Kalai, A. (2016). Man is to Computer Programmer as
Woman is to Homemaker? Debiasing Word Embedding [Paper presentation]. 30th Conference on Neural
Information Processing Systems, Spain. https://doi.org/arxiv.org/pdf/1607.06520.pdf
Botsman, R. (2017). Who Can You Trust?: How Technology Brought Us Together–and Why It Could Drive Us
Apart. Penguin. https://play.google.com/store/books/details?id=7cGWDgAAQBAJ
Brantingham, P. J., Valasik, M., & Mohler, G. O. (2018). Does Predictive Policing Lead to Biased Arrests?
Results From a Randomized Controlled Trial. Statistics and Public Policy, 5(1), 1-6.
http://doi.org/10.1080/2330443x.2018.1438940
Bravo Orellana, E. R. (2015). Deskilling, Up-skilling or Reskilling? Effects of Automation in Information
Systems Context. 21st Americas Conference on Information Systems, Puerto Rico.
https://pdfs.semanticscholar.org/0d46/4ed08abb89434c063ea5145564598ac46c8b.pdf?_ga=2.268117580.396
720421.1584246117-1624914399.1584246117
Brundrett, S. (2014). Industrial Analysis of Autonomous Mine Haul Truck Commercialisation. [Masters’ thesis,
Beedie School of Business-Segal Graduate School], Summit. https://summit.sfu.ca/item/14425
Buolamwini, J., & Gebru, T. (2018). Gender Shades: Intersectional Accuracy Disparities in Commercial Gender
Classification. 1st Conference on Fairness, Accountability, and Transparency: Proceedings of Machine
Learning Research, Atlanta. http://proceedings.mlr.press/v81/buolamwini18a/buolamwini18a.pdf
Casner, S. M., Hutchins, E. L., & Norman, D. (2016). The challenges of partially automated driving.
Communications of the ACM, 59(5), 70-77. http://doi.org/10.1145/2830565
Caterpillar. (n.d.-a). Autonomous Mining: Improving Safety and Increasing Productivity.
https://www.cat.com/en_US/by-industry/mining/articles/improving-safety-and-productivity.html
Caterpillar. (n.d.-b). Avoidance zone: building virtual fences to enhance safety. https://www.cat.com/en_US/by-
industry/mining/articles/avoidance-zones.html
Caterpillar. (n.d.-c). CAT® Object Detection. Caterpillar. https://www.cat.com/en_US/by-
industry/mining/surface-mining/surface-technology/detect/detect-for-object-detection.html
Caterpillar. (2013). Cat® Command for Hauling. Caterpillar.
http://s7d2.scene7.com/is/content/Caterpillar/C10338825
Caterpillar. (n.d.-e). Cat® command for hauling users report gains. Caterpillar.
https://www.cat.com/en_US/articles/customer-stories/mining/viewpoint/autonomous-haulage-gains.html
Chandler, D. (2014). Beyond neoliberalism: resilience, the new art of governing complexity. Resilience, 2(1),
47-63. http://doi.org/10.1080/21693293.2013.878544
Christoffersen, K., & Woods, D. D. (2002). How to Make Automated Systems Team Players. In Salas, E. (Ed.),
Advances in Human Performance and Cognitive Engineering Research (Vol. 2, pp. 1-12). Emerald Group
Publishing Limited. http://csel.eng.ohio-state.edu/productions/xcta/downloads/automation_team_players.pdf

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 29

Cilliers, P., & Presier (2010). Complexity, Difference and Identity: An Ethical Perspective. Springer.
https://books.google.com.au/books?id=T0_jyLRD_c4C&printsec=frontcover&source=gbs_ge_summary_r&c
ad=0#v=onepage&q&f=false
Coppola, G. (2018). Supplier Claims Uber Disabled Its Safety Tech in Volvo That Caused Pedestrian Fatality.
Insurance Journal. https://www.insurancejournal.com/news/national/2018/03/27/484440.htm
Creswell, J. W., & Creswell, J. D. (2017). Research Design: Qualitative, Quantitative, and Mixed Methods
Approaches (5th ed.). SAGE.
https://play.google.com/store/books/details/John_W_Creswell_Research_Design?id=335ZDwAAQBAJ&hl=
en_US
Cummings, M. L., Kilgore, R. M., Wang, E., Tijerina, L., & Kochhar, D. S. (2016). Effects of Single Versus
Multiple Warnings on Driver Performance. Human Factors: The Journal of the Human Factors and
Ergonomics Society, 49(6), 1097-1106. http://doi.org/10.1518/001872007x249956
Dahlstrom, N., Dekker, S. W. A., van Winsen, R., & Nyce, J. (2009). Fidelity and validity of simulator training.
Theoretical Issues in Ergonomics Science, 10(4), 305-314. http://doi.org/10.1080/14639220802368864
De Boer, R., & Dekker, S. W. A. (2017). Models of Automation Surprise: Results of a Field Survey in Aviation.
Special Issue Aviation Safety, 3(3), 1-11. http://doi.org/10.3390/safety3030020
de Visser, E. J., Pak, R., & Shaw, T. H. (2018). From 'automation' to 'autonomy': the importance of trust repair in
human-machine interaction. Ergonomics, 61(1) 1-33. http://doi.org/10.1080/00140139.2018.1457725
de Vries, L. (2017). Work as Done? Understanding the Practice of Sociotechnical Work in the Maritime
Domain. Journal of Cognitive Engineering and Decision Making, 11(3), 270-295.
http://doi.org/10.1177/1555343417707664
de Winter, J. C. F., & Dodou, D. (2011). Why the Fitts list has persisted throughout the history of function
allocation. Cognition, Technology & Work, 16(1), 1-11. http://doi.org/10.1007/s10111-011-0188-1
Dekker, S. (2019). Foundations of Safety Science: A Centure of Understanding Accidents and Disasters. CRC
Press. https://play.google.com/books/reader?id=dwWSDwAAQBAJ&printsec
Dekker, S. (2003). Failure to adapt or adaptations that fail: contrasting models on procedures and safety. Applied
Ergonomics, 34(3), 233-238. http://doi.org/10.1016/s0003-6870(03)00031-0
Dekker, S. (2004). To engineer is to err. In Sandom, C. & Harvey, R. (Eds.), Human Factors for Engineers (1st
ed., pp. 137-150). The Institution of Engineering and Technology. http://doi.org/10.1049/PBNS032E
Dekker, S. (2010). We have Newton on a retainer: Reductionism when we need systems thinking. The Joint
Commission Journal on Quality and Patient Safety, 36(4), 152–163. http://doi.org/10.1016/s1553-
7250(10)36024-7
Dekker, S. (2014a). Field Guide to Understanding 'Human Error'. CRC Press.
https://ebookcentral.proquest.com/lib/curtin/reader.action?docID=1825729
Dekker, S. (2014b). Safety differently: humans factors for a new era. CRC Press.
https://books.google.com.au/books/about/Safety_Differently.html?id=MefMAwAAQBAJ&printsec=frontcov
er&source=kp_read_button&redir_esc=y#v=onepage&q&f=false
Dekker, S., Bergström, J., Amer-Wåhlin, I., & Cilliers, P. (2012). Complicated, complex, and compliant: best
practice in obstetrics. Cognition, Technology & Work, 15(2), 189–195. http://doi.org/10.1007/s10111-011-
0211-6
Dekker, S., Hollnagel, E., Woods, D. D., & Cook, R. (2008). Resilience Engineering: New directions for
measuring and maintaining safety in complex system (Final Report). Lund University School of Aviation.
https://pdfs.semanticscholar.org/a0d3/9cc66adc64e297048a32b71aeee209a451af.pdf?_ga=2.1134067.396720
421.1584246117-1624914399.1584246117

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 30

Dekker, S., & Woods, D. D. (2002a). MABA-MABA or Abracadabra? Progress on Human–Automation Co-
ordination. Cognition, Technology & Work, 4(4), 240–244. http://doi.org/10.1007/s101110200022
Department of Mines and Petroluem. (2014a). Fatal accidents in the Western Australian mining industry 2000-
2012: What lessons can we learn? Investigation Services and Resources Safety Division.
https://www.dmp.wa.gov.au/Documents/Safety/MSH_R_FatalAccidents200012.pdf
Department of Mines and Petroluem. (2014b). Mines Safety Bulletin: Seeking safe mobile autonomous
equipment systems (Mines Safety Bulletin No. 110). https://www.dmp.wa.gov.au/Documents/Safety/SRS-
Publications-Mining_and_Explorations-Safety_Bulletin_114.pdf
Department of Mines and Petroluem. (2015a). Code of Practice: Safe mobile autonomous mining in Western
Australia.
https://www.dmp.wa.gov.au/Documents/Safety/MSH_COP_SafeMobileAutonomousMiningWA.pdfhttp://w
ww.dmp.wa.gov.au/Documents/Safety/MSH_COP_SafeMobileAutonomousMiningWA.pdf
Department of Mines and Petroluem. (2015b). Haul truck driving - mine safety matters [Pamphlet].
http://www.dmp.wa.gov.au/Documents/Safety/MSH_MSM_P_TruckDriving.pdf
Department of Mines and Petroluem. (2015c). Signficant Incident Report: Collision between an autonomous
haul truck and manned water cart (Significant Incident Report No. 226).
http://www.dmp.wa.gov.au/Documents/Safety
MS_SIR_226_Collision_between_an_autonomous_haul_truck_and_manned_water_cart.pdf
Dietvorst, B. J., Simmons, J. P., & Massey, C. (2016). Overcoming Algorithm Aversion: People Will Use
Imperfect Algorithms If They Can (Even Slightly) Modify Them. Management Science, 64(3), 1155-1170.
http://doi.org/10.1287/mnsc.2016.2643
Dixon, S. R., Wickens, C. D., & McCarley, J. S. (2007). On the independence of compliance and reliance: are
automation false alarms worse than misses? Human Factors: The Journal of Human Factors and Ergonomics
Society, 49(4), 564-572. http://doi.org/10.1518/001872007X215656
Domingos, P. (2015). The master algorithm: How the quest for the ultimate learning machine will remake our
world. Penguin.
https://books.google.com.au/books/about/The_Master_Algorithm.html?id=pjRkCQAAQBAJ&printsec=front
cover&source=kp_read_button&redir_esc=y#v=onepage&q&f=false
Dominiczak, J., & Khansa, L. (2018). Principles of Automation for Patient Safety in Intensive Care: Learning
From Aviation. Joint Commission Journal Quality Patient Safety, 44(6), 366-371.
http://doi.org/10.1016/j.jcjq.2017.11.008
Dressel, J., & Farid, H. (2018). The accuracy, fairness, and limits of predicting recidivism. Science Advances,
4(1), 1-5. http://doi.org/10.1126/sciadv.aao5580
Drury, C. G., Porter, W. L., & Dempsey, P. G. (2012). Patterns in Mining Haul Truck Accidents [Paper
presentation]. Proceedings of the Human Factors and Ergonomics Society 56th Annual Meeting, 56(1), 2011-
2015. http://doi.org/10.1177/1071181312561420
Earley, S. (2016). There Is No AI Without IA. IEEE Computer Society, 18(3), 58-64.
http://doi.org/10.1109/MITP.2016.43
Edwards, E. (1977). Automation in civil transport aircraft. Applied Ergonomics, 8(4), 194–198.
http://doi/org10.1016/0003-6870(77)90 3-6
Endsley, M. R. (2016a). Designing for Situation Awareness: An Approach to User-Centered Design (2nd ed.).
CRC Press. https://www-taylorfrancis-com.dbgw.lis.curtin.edu.au/books/9781420063585
Endsley, M. R. (2016b). From Here to Autonomy: Lessons Learned from Human-Automation Research. Human
Factors: The Journal of the Human Factors and Ergonomics Society, 59(1), 5-27.
http://doi.org/10.1177/0018720816681350

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 31

Endsley, M. R. (2017). Level of Automation Forms a Key Aspect of Autonomy Design. Journal of Cognitive
Engineering and Decision Making, 12(1), 29-34. http://doi.org/10.1177/1555343417723432
Endsley, M. R. (2018). Situation Awareness in Future Autonomous Vehicles: Beware of the Unexpected [Paper
presentation]. In: Bagnara S., Tartaglia R., Albolino S., Alexander T., Fujita Y. (Eds), 20th Congress of the
International Ergonomics Association (pp. 303-309), Springer. http://doi.org/10.1007/978-3-319-96071-5_32
Endsley, M. R., & Kiris, E. O. (1995). The Out-of-the-Loop Performance Problem and Level of Control in
Automation. Human Factors: The Journal of the Human Factors and Ergonomics Society, 37(2), 381-394.
http://doi.org/10.1518/001872095779064555
Engle, R. W. (2016). Working Memory Capacity as Executive Attention. Current Directions in Psychological
Science, 11(1), 19-23. http://doi.org/10.1111/1467-8721.00160
Ernst and Young. (2019). The Future of Work The economic implications of technology and digital mining: A
Report for the Minerals Council of Australia (Report No. 190214). Minerals Council of Australia.
http://minerals.org.au/sites/default/files/190214%20The%20Future%20of%20Work%20The%20economic%
20implications%20of%20technology%20and%20digital%20mining.pdf
Eykholt, K., Evtimov, I., Fernandes, E., Li, B., Rahmati, A., Xiao, C., . . . Song, D. (2018). Robust Physical-
World Attacks on Deep Learning Visual Classification [Paper presentation]. 2018 IEEE/CVF Conference on
Computer Vision and Pattern Recognition. http://doi.org/10.1109/CVPR.2018.00175
Feldhütter, A., Härtwig, N., Kurpiers, C., Hernandez, J. M., & Bengler, K. (2019). Effect on Mode Awareness
When Changing from Conditionally to Partially Automated Driving. In: Bagnara S., Tartaglia R., Albolino
S., Alexander T., Fujita Y. (Eds). 20th Congress of the International Ergonomics Association (pp. 314-324),
Springer. https://doi.org/10.1007/978-3-319-96074-6_34
Ferris, T., Sarter, N. B., & Wickens, C. D. (2010). Cockpit Automation: Still trying to catch up…. In Salas, E., &
Maurino, D. (Eds.), Human Factors in Aviation (2nd ed., pp. 479-503). Academic Press.
https://doi.org/10.1016/B978-0-12-374518-7.00015-8
Fisher, B. S., & Schnittger, S. (2012). Autonomous and Remote Operation Technologies in the Mining Industry:
Benefits and Costs (BAE Research Report 12.1). BAEcomonics. http://www.baeconomics.com.au/wp-
content/uploads/2010/01/Mining-innovation-5Feb12.pdf
Fridman, L. (2018). Human-Centered Autonomous Vehicle Systems: Principles of Effective Shared Autonomy.
https://arxiv.org/pdf/1810.01835.pdf
Fridman, L., Brown, D. E., Glazer, M., Angell, W., Dodd, S., Jenik, B., . . . Reimer, B. (2018). MIT
Autonomous Vehicle Technology Study: Large-Scale Deep Learning Based Analysis of Driver Behavior and
Interaction with Automation. IEEE Access, 7, 102021-102038.
http://doi/org//10.1109/ACCESS.2019.2926040
Fridman, L., Ding, L., Jenik, B., & Reimer, B. (2018). Arguing Machines: Human Supervision of Black Box AI
Systems That Make Life-Critical Decisions. https://arxiv.org/pdf/1710.04459v2.pdf
Frimpong, S., Changirwa, R., & Szymanski, J. (2003). Simulation of Automated Dump Trucks for Large Scale
Surface Mining Operations. International Journal of Surface Mining, Reclamation and Environment, 17(3),
183-195. http://doi.org/10.1076/ijsm.17.3.183.14770
Frohm, J., Lindström, V., Winroth, M., & Stahre, J. (2006). The Industry's View on Automation in
Manufacturing. IFAC Proceedings Volumes, 39(4), 453-458. http://doi.org/10.3182/20060522-3-fr-
2904.00073
Funk, K., Lyall, B., Wilson, J., Vint, R., Niemczyk, M., Suroteguh, C., & Owen, G. (2009). Flight Deck
Automation Issues. The International Journal of Aviation Psychology, 9(2), 109–123.
http://doi.org/10.1207/s15327108ijap0902_2

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 32

Giacomin, J. (2015). What Is Human Centred Design? The Design Journal, 17(4), 606-623.
http://doi.org/10.2752/175630614x14056185480186
Glover, M. (2016, Oct 4-6). Caterpillar’s Autonomous Journey - The Argument for Autonomy [Paper
presentation]. SAE 2016 Commercial Vehicle Engineering Congress, USA. https://doi.org/10.4271/2016-01-
8005.
Goel, A. (2016). What Tech Will it Take to Put Self-Driving Cars on the Road? Engineering.com.
https://www.engineering.com/DesignerEdge/DesignerEdgeArticles/ArticleID/13270/What-Tech-Will-it-
Take-to-Put-Self-Driving-Cars-on-the-Road.aspx
Goteman, y., & Dekker, S. (2007). Flight Crew Callouts and Aircraft Automation Modes An Observational
Study of Task Shedding. International Journal of Applied Aviation Studies, 6(2), 235-248.
http://www.humanfactors.lth.se/fileadmin/lusa/Sidney_Dekker/articles/2006/ModecalloutsIJAAS.pdf
Gschwandtner, M., Pre, W., & Uhl, A. (2010). Track Detection for Autonomous Trains. In Bebis, G., Boyle, R.,
Parvin, B., Koracin, D., Chung, R., Hammound, R., Hussain, M., Kar-Han, T., Crawfis, R., Thalmann, D.,
Kao, D., & Avila, L. (Eds.), Lecture Notes in Computer Science (Vol. 6455, pp. 19-28).
https://doi.org/10.1007/978-3-642-17277-9_3
Gunning, D. (2016). Explainable Artificial Intelligence (XAI) [PowerPoint slides]. DARPA.
https://www.darpa.mil/program/explainable-artificial-intelligence
Hamada, T., & Saito, S. (2018). Autonomous Haulage System for Mining Rationalisation. Hitachi Review,
67(1), 87-92. http://www.hitachi.com/rev/archive/2018/r2018_01/pdf/P087-092_R1a07.pdf
Hancock, A. P., Jagacinski, J. R., Parasuraman, R., Wickens, C. D., Wilson, F. G., & Kaber, B. D. (2013).
Human-Automation Interaction Research: Past, Present, and Future. Ergonomics in Design: The Quarterly of
Human Factors Applications, 21(2), 9-14. http://doi.org/10.1177/1064804613477099
Hansen, A. [Andrew Hansen] (2020, Jan 31). Autonomous Dump Trucks going around a broken down truck
[Video]. YouTube. https://www.youtube.com/watch?v=GEu2Ijvkl0U
Hebbar, A. (2017, Nov 3-5). Augmented intelligence: Enhancing human capabilities [Paper presentation]. 2017
Third International Conference on Research in Computational Intelligence and Communication Networks
(ICRCICN), India. http://doi.org/10.1109/ICRCICN.2017.8234515
Held, D., Levinson, J., & Thrun, S. (2012, May 14-18). A Probalistic Framework for Car Detection in Images
using Context and Scale [Paper presentation]. IEEE International Conference on Robotics and Automation,
USA. http://doi.org/10.1109/ICRA.2012.6224722
Hew, P. C. (2016). Detecting Occurrences of the “Substitution Myth”: A Systems Engineering Template for
Modeling the Supervision of Automation. Journal of Cognitive Engineering and Decision Making, 11(2),
184-199. http://doi.org/10.1177/1555343416674422
Hitachi. (2015, Feb 19). Hitachi dump trucks Autonomous Haulage Solution – AHS [Video]. YouTube.
https://www.youtube.com/watch?v=c9_Os6Ha-Gk&t=84s
Hoff, K. A., & Bashir, M. (2015). Trust in automation: integrating empirical evidence on factors that influence
trust. Human Factors, 57(3), 407-434. http://doi.org/10.1177/0018720814547570
Hollnagel, E., Wears, R. L., & Braithwaite, J. (2015). From Safety-I to Safety-II: A White Paper. National
Library of Congress. https://www.england.nhs.uk/signuptosafety/wp-content/uploads/sites/16/2015/10/safety-
1-safety-2-whte-papr.pdf
Holroyd, J. (2015). Implicit bias, awareness and imperfect cognitions. Consciousness and Cognition, 33, 511-
523. http://doi.org/10.1016/j.concog.2014.08.024

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 33

Horberry, T. (2012). The Health and Safety Benefits of New Technologies in Mining: A Review and Strategy for
Designing and Deploying Effective User-Centred Systems. Minerals, 2(4), 417-425.
http://doi.org/10.3390/min2040417
Horberry, T., Lynas, D., Franks, D., Barnes, R., & Brereton, D. (2011, 22-23 November). Brave new mine –
examining the human factors implications of automation and remote operation in mining [Paper
presentation]. Second International Future Mining Conference, Australia. https://ausimm.com/product/brave-
new-mine-examining-the-human-factors-implications-of-automation-and-remote-operation-in-mining/
Hurley, M., & Adebayo, J. (2016). Credit scoring in the era of big data. Yale Journal of Law and Technology,
18(1). https://digitalcommons.law.yale.edu/yjolt/vol18/iss1/5
Inagaki, T. (2003). Adaptive Automation: Sharing and Trading of Control. In Hollnagel, E. (Ed.), Handbook of
Cognitive Task Design (pp. 147–169). CRC Press. http://doi.org/10.1299/jsmetld.2001.10.79
International Organization for Standardization. (2019). Earth-moving machinery and mining — Autonomous and
semi-autonomous machine system safety (ISO 17757:2019). https://www.iso.org/standard/76126.html
Ito, J., & Howe, J. (2016). Whiplash: How to survive our faster future. Grand Central Publishing.
Jamasmie, C. (2019, March 19). BHP blames heavy rains for autonomous trucks crash.
http://www.mining.com/bhp-blames-heavy-rains-autonomous-trucks-crash/
Körber, M., Baseler, E., & Bengler, K. (2018). Introduction matters: Manipulating trust in automation and
reliance in automated driving. Applied Ergonomics, 66, 18-31. http://doi.org/10.1016/j.apergo.2017.07.006
Kyriakidis, M., de Winter, J. C. F., Stanton, N., Bellet, T., van Arem, B., Brookhuis, K., . . . Happee, R. (2017).
A human factors perspective on automated driving. Theoretical Issues in Ergonomics Science, 20(3), 223-
249. http://doi.org/10.1080/1463922x.2017.1293187
Lake, B. M., Lee, C., Glass, J. R., & Tenenbaum, J. B. (2014, July 23–26). One-shot learning of generative
speech concepts. Proceedings of the Annual Meeting of the Cognitive Science Society, Quebec City, QC,
Canada. https://cloudfront.escholarship.org/dist/prd/content/qt3xf2n3vc/qt3xf2n3vc.pdf
Lake, B. M., Salakhutdinov, R., & Tenenbaum, J. B. (2015). Human-level concept learning through probabilistic
program induction. Science, 350, 1332-1338. http://doi.org/10.1126/science.aab3050
Lake, B. M., Ullman, T. D., Tenenbaum, J. B., & Gershman, S. J. (2016). Building machines that learn and think
like people. Behavioral and Brain Sciences, 40, 1-101. http://doi.org/10.1017/S0140525X16001837
Lee, J. D., & See, K. A. (2004). Trust in Automation: Designing for Appropriate Reliance. Human Factors: The
Journal of the Human Factors and Ergonomics Society, 46(1), 50–80.
http://doi.org/10.1518/hfes.46.1.50_30392
Li, H., Wickens, C. D., Sarter, N. B., & Sebok, A. (2014). Stages and levels of automation in support of space
teleoperations. Human Factors, 56(6), 1050-1061. http://doi.org/10.1177/0018720814522830
Lützhöft, M. H., & Dekker, S. W. A. (2002). On Your Watch: Automation on the Bridge. Journal of Navigation,
55(1), 83-96. http://doi.org/10.1017/S0373463301001588
Lynas, D., & Horberry, T. (2010, Oct 31 - Nov 3). Exploring the Human Factors Challenges of Automated
Mining Equipment. 46th Annual Human Factors and Ergonomics Society of Australia Conference, Australia.
Lynas, D., & Horberry, T. (2011). A review of Australian human factors research and stakeholder opinions
regarding mines of the future. Paper presented at the HFESA 47th Annual Conference 2011. Ergonomics
Australia - Special Edition, Australia.
McCarthy, I. P., Rakotobe-Joel, T., & Frizelle, G. (2000). Complex systems theory: implications and promises
for manufacturing organisations. International Journal of Manufacturing Technology and Management, 2(1-
7), 559-579. http://doi.org/10.1504/ijmtm.2000.001365

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 34

McFarland, M. (2017). Google's AI just beat the world's best Go player.

https://money.cnn.com/2017/05/25/technology/alphago-china-ai/index.html
McKinnon, S. (2019). Fortescue Metals Group auto haul truck crash Christmas Creek ‘no failure’ of system.
https://thewest.com.au/business/mining/fortescue-metals-group-auto-haul-truck-crash-christmas-creek-no-
failure-of-system-ng-b881104957z
Merritt, S. M., Lee, D., Unnerstall, J. L., & Huber, K. (2015). Are well-calibrated users effective users?
Associations between calibration of trust and performance on an automation-aided task. Human Factors,
57(1), 34-47. http://doi.org/10.1177/0018720814561675
Miller, C. A., & Parasuraman, R. (2007). Designing for flexible interaction between humans and automation:
delegation interfaces for supervisory control. Human Factors, 49(1), 57-75.
http://doi.org/10.1518/001872007779598037
MIT Sloan CIO Symposium Videos. (2017, June 13). 2017-04 Putting AI to Work [Video]. YouTube.
https://www.youtube.com/watch?v=kDvn2eOOlz4
Mitchell, R. (2018). Driverless cars are growing in number, but makers don't want to reveal how they sometimes
fail. https://www.latimes.com/business/autos/la-fi-hy-driverless-data-20180430-story.html
Mols, B., & Vergunst, N. (2018). Hallo Robot: Meet Your New Workmate and Friend.
https://popsciencebooks.blogspot.com/2018/11/hallo-robot-bennie-mols-and-nieske.html
Murphy, M. (2016). The Journey of Using Autonomous Machines in Mining.
http://www.aaes.org/sites/default/files/Murphy_Convocation2017.pdf.
National Transportation Safety Board. (2017). Collision Between a Car Operating With Automated Vehicle
Control Systems and a Tractor-Semitrailer Truck Near Williston, Florida, May 7, 2016.
National Transportation Safety Board. (2018). Preliminary Report Highway: HWY18MH010.
https://www.ntsb.gov/investigations/AccidentReports/Reports/HWY18MH010-prelim.pdf
Norman, D. (2013). The Design of Everyday Things. http://www.nixdell.com/classes/HCI-and-Design-Spring-
2017/The-Design-of-Everyday-Things-Revised-and-Expanded-Edition.pdf
Noy, I. Y., Shinar, D., & Horrey, W. J. (2018). Automated driving: Safety blind spots. Safety Science, 102, 68-
78. http://doi.org/10.1016/j.ssci.2017.07.018
Ockerman, J., & Pritchett, A. R. (2002). Impact of Contextual Information on Automation Brittleness. Human
Factors and Ergonomics Society 46th Annual Meeting, USA. http://doi.org/10.1177/154193120204600335
Palmer, J. (2019). James Palmer BMA Asset President speaks at Bowen Basin Mining Club 2019. BHP.
https://www.bhp.com/media-and-insights/reports-and-presentations/2019/07/james-palmer-bma-asset-
president-speaks-at-bowen-basin-mining-club-2019
Panetta, K. (2019). 5 Trends Appear on the Gartner Hype Cycle for Emerging Technologies, 2019. Gartner.
https://www.gartner.com/smarterwithgartner/5-trends-appear-on-the-gartner-hype-cycle-for-emerging-
technologies-2019/
Parasuraman, R., & Riley, V. (1997). Humans and Automation: Use, Misuse, Disuse, Abuse. Human Factors:
The Journal of the Human Factors and Ergonomics Society, , 39(2), 230–253.
http://doi.org/10.1518/001872097778543886
Parasuraman, R., Sheridan, T. B., & Wickens, C. D. (2000). A Model for Types and Levels of Human
Interaction with Automation. IEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and
Humans, 30(3), 286-297. http://doi.org.10.1109/3468.844354
Payre, W., Cestac, J., & Delhomme, P. (2016). Fully Automated Driving: Impact of Trust and Practice on
Manual Control Recovery. Human Factors, 58(2), 229-241. http://doi.org/10.1177/0018720815612319

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 35

Perrow, C. (1984). Normal Accidents: Living with High-Risk Technologies. Princeton University Press.
https://books.google.com.au/books?hl=en&lr=&id=g66J6Vzq6EYC&oi=fnd&pg=PR5&dq=normal+accident
s+perrow&ots=Wr_AzoHouu&sig=Dw4nR9eJUoSCuvlfFIaeWNA1Egs#v=onepage&q=normal%20accident
s%20perrow&f=false
Pettersen, K. A., & Schulman, P. R. (2016). Drift, adaptation, resilience and reliability: Toward an empirical
clarification. Safety Science. http://doi.org/10.1016/j.ssci.2016.03.004
Prechelt, L. (2012). Early Stopping — But When? Neural Networks: Tricks of the Trade, 53–67.
http://doi.org/10.1007/978-3-642-35289-8_5
Prewett, M. S., Johnson, R. C., Saboe, K. N., Elliott, L. R., & Coovert, M. D. (2010). Managing workload in
human–robot interaction: A review of empirical studies. Computers in Human Behavior, 26(5), 840-856.
http://doi.org/10.1016/j.chb.2010.03.010
Pritchett, A. R., Kim, S. Y., & Feigh, K. (2013). Measuring Human-Automation Function Allocation. Journal of
Cognitive Engineering and Decision Making, 8(1), 52-77. http://doi.org/10.1177/1555343413490166
Protzman, C., Whiton, F., Kerpchar, J., Lewandowski, C., Stenberg, S., & Grounds, P. (2016). Harada Method
By Norman Bodek. In The Lean Practitioner's Field Book: Proven, Practical, Profitable and Powerful
Techniques for Making Lean Really Work.
http://ebookcentral.proquest.com/lib/curtin/detail.action?docID=4709745.
Rankin, A., Woltjer, R., & Field, J. (2016). Sensemaking following surprise in the cockpit—a re-framing
problem. Cognition, Technology & Work, 18(4), 623-642. http://doi.org/10.1007/s10111-016-0390-2
Rasmussen, J., & Vincente, K. (1989). Coping with human errors through system design: implications for
ecological interface design. International Journal of Human-Computer Studies, 31(5), 517-534.
http://https://doi.org/10.1016/0020-7373(89)90014-X
Reason, J. (1990). Latent errors and systems disasters. In Human Error (pp. 173-216). Cambridge University
Press.
https://ebookcentral.proquest.com/lib/curtin/detail.action?docID=691788
Rio Tinto. (n.d.). How did the world's biggest robot end up here? https://www.riotinto.com/en/news/stories/how-
did-worlds-biggest-robot
Ross, C., & Swetlitz, I. (2018). IBM’s Watson supercomputer recommended ‘unsafe and incorrect’ cancer
treatments, internal documents show. https://www.statnews.com/2018/07/25/ibm-watson-recommended-
unsafe-incorrect-treatments/
Rousseau, J. (2015). Why Detroit Needs User-Centered Design. https://www.fastcompany.com/3054072/why-
detroit-needs-user-centered-design
Salas, E., Jentsch, F., & Maurino, D. (2010). Human Factors in Aviation (2nd ed.).
http://www.sciencedirect.com.dbgw.lis.curtin.edu.au/science/book/9780123745187
Sarter, N. B. (2008). Investigating mode errors on automated flight decks: illustrating the problem-driven,
cumulative, and interdisciplinary nature of human factors research. Human Factors, 50(3), 506-510.
http://doi.org/10.1518/001872008X312233
Sarter, N. B., Mumaw, R. J., & Wickens, C. D. (2007). Pilots' monitoring strategies and performance on
automated flight decks: an empirical study combining behavioral and eye-tracking data. Human Factors,
49(3), 347-357. http://doi.org/10.1518/001872007X196685
Sarter, N. B., & Woods, D. D. (1994). Pilot Interaction With Cockpit Automation II: An Experimental Study of
Pilots' Model and Awareness of the Flight Management System. The International Journal of Aviation
Psychology, 4(1), 1-28. http://doi.org/10.1207/s15327108ijap0401_1

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 36

Sarter, N. B., & Woods, D. D. (1995). How in the World Did We Ever Get into That Mode? Mode Error and
Awareness in Supervisory Control. Human Factors, 37(1), 5-19.
http://doi.org/10.1518/001872095779049516
Sarter, N. B., Woods, D. D., & Billings, C. E. (1997). Automation surprises. In Salvendy, G. (Ed.), Handbook of
Human Factors & Ergonomics (Vol. 2nd ed., pp. 1926-1943). John Wiley & Sons.
Schmitt, K. (2012). Automations influence on nuclear power plants: a look at three accidents and how
automation played a role. Work, 41(1), 4545-4551. http://doi.org/10.3233/WOR-2012-0035-4545
Sieck, W., Klein, G., Peluso, D. A., Smith, J. L., & Harris-Thompson, D. (2007). A Model of Sensemaking
(Technical Report 1200). https://www.researchgate.net/publication/266217173_A_Model_of_Sensemaking
Skeem, J. L., & Lowenkamp, C. (2016). Risk, Race, & Recidivism: Predictive Bias and Disparate Impact.
Criminology, 54(4), 680-712. http://doi.org/10.1111/1745-9125.12123
Sklar, A. E., & Sarter, N. B. (1999). Tactile Feedback in Support of Attention Allocation and Human-
Automation Coordination in Event-Driven Domains. Human Factors: The Journal of Human Factors and
Ergonomics Society, 41(4), 543-552. http://doi.org/10.1518/001872099779656716
SlashGear. (2017, 11 Sep). 2019 Audi A8 Level 3 self-driving real world test [Video]. YouTube.
https://www.youtube.com/watch?v=WsiUwq_M8lE
Srinivasan, B. N., & Mukherjee, D. (2018). Agile teams as complex adaptive systems (CAS). International
Journal of Information Technology, 10(3), 367-378. http://doi.org/10.1007/s41870-018-0122-3
Stanton, N. A., Chambers, P. R. G., & Piggott, J. (2001). Situational awareness and safety. Safety Science, 39(3),
189-204. http://https://doi.org/10.1016/S0925-7535(01)00010-8
Stensson, P., & Jansson, A. (2014). Autonomous technology - sources of confusion: a model for explanation and
prediction of conceptual shifts. Ergonomics, 57(3), 455-470. http://doi.org/ 10.1080/00140139.2013.858777
Strand, N., Nilsson, J., Karlsson, I. C. M., & Nilsson, L. (2014). Semi-automated versus highly automated
driving in critical situations caused by automation failures. Transportation Research Part F: Traffic
Psychology and Behaviour, 27, 218-228. http://doi.org/10.1016/j.trf.2014.04.005
Strand, N., Stave, C., & Ihlström, J. (2018, September 17-21). A case-study on drivers’ mental model of partial
driving automation [Paper presentation]. 25th ITS World Congress, Denmark.
https://www.researchgate.net/publication/328476996_A_case
study_on_drivers'_mental_model_of_partial_driving_automation
Summers, A. E. (2003). Introduction to layers of protection analysis. Journal of Hazardous Materials, 104(1-3),
163-168. http://10.1016/s0304-3894(03)00242-5
Tech Light. (2016). Toyota is securing efficiency by replacing robots with humans. Investor Spot.
http://inventorspot.com/articles/toyota-securing-efficiency-replacing-robots-humans
Teichman, A., Levinson, J., & Thrun, S. (2011, 9-13 May). Towards 3D Object Recognition via Classification of
Arbitrary Object Tracks. 2011 IEEE International Conference on Robotics and Automation, China.
http://10.1109/ICRA.2011.5979636
The Wheel Network. (2016, 11 April). Lexus Takumi: Master craftsman at the Kentucky plant [Video].
YouTube. https://www.youtube.com/watch?v=yH5_McdFWl4
Today Tonight. (2018, February 28). Huge mining trucks with NO DRIVERS and the massive machines operated
by an Xbox controller [Video]. YouTube. https://www.facebook.com/TodayTonight/videos/huge-mining-
trucks-with-no-drivers-and-the-massive-machines-operated-by-an-xbox-/1706979276008188/
Trudell, C., Hagiwara, Y., & Jie, M. (2014). Humans Replacing Robots Herald Toyota’s Vision of Future.
https://www.bloomberg.com/news/articles/2014-04-06/humans-replacing-robots-herald-toyota-s-vision-of-
future
https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 37

Victor, T. W., Tivesten, E., Gustavsson, P., Johansson, J., Sangberg, F., & Ljung Aust, M. (2018). Automation
Expectation Mismatch: Incorrect Prediction Despite Eyes on Threat and Hands on Wheel. Human Factors,
60(8), 1095-1116. http://doi.org/10.1177/0018720818788164
von Hippel, E. (2005). Democratizing innovation: The evolving phenomenon of user innovation. Journal für
Betriebswirtschaft, 55(1), 63-78. http://do.org/10.1007/s11301-004-0002-8
Vul, E., Goodman, N., Griffiths, T. L., & Tenenbaum, J. B. (2014). One and done? Optimal decisions from very
few samples. Cogn Sci, 38(4), 599-637. http://doi.org/10.1111/cogs.12101
Wakabayashi, D. (2018, May 24). Emergency Braking Was Disabled When Self-Driving Uber Killed Woman,
Report Says. https://www.nytimes.com/2018/05/24/technology/uber-autonomous-car-ntsb-investigation.html
Walker, J. (2016). Google's AI can now caption images almost as well as humans.
http://www.digitaljournal.com/tech-and-science/technology/google-s-ai-now-captions-images-with-94-
accuracy/article/475547
Wears, R. L., Hollnagel, E., & Braithwaite, J. (2015). Resilient health care. The Resilience of Everyday Clinical
Work. https://doi.org/10.1201/9781315605739
Weber, D. E., & Dekker, S. W. A. (2017). Assessing the sharp end reflections on pilot performance assessment
in the light of Safety Differently. Theoretical Issues in Ergonomics Science, 18(1), 59-78.
http://doi.org/10.1080/1463922X.2016.1149253
Wessel, G., Altendorf, E., Schreck, C., Canpolat, Y., & Flemisch, F. (2019). Cooperation and the Role of
Autonomy in Automated Driving. In Waschl, H., Kolmanovsky, I., Willems F. (eds), Control Strategies for
Advanced Driver Assistance Systems and Autonomous Driving Functions, Lecture Notes in Controland
Information Sciences, (Vol 476, pp. 1-27). Springer. http://doi.org/10.1007/978-3-319-91569-2_1
Wickens, C. D. (2008). Multiple resources and mental workload. Human Factors, 50(3), 449-455.
http://doi.org/10.1518/001872008X288394
Wickens, C. D., Hollands, J. G., Banbury, S., & Parasuraman, R. (2016). Engineering Psychology and Human
Performance. https://books.google.com.au/books https://books.google.com.au/books
Wickens, C. D., Rice, S., Keller, D., Hutchins, S., Hughes, J., & Clayton, K. (2009). False alerts in air traffic
control conflict alerting system: is there a "cry wolf" effect? Human Factors, 51(4), 446-462.
http://doi.org/10.1177/0018720809344720
Wiener, E. L. (1989). Human Factors of Advanced Technology (“Glass Cockpit”) Transport Aircraft (NASA
Contractor Report 177528). National Aeronautics and Space Administration.
https://ntrs.nasa.gov/search.jsp?R=19890016609
Willey, R. J. (2014). Layer of Protection Analysis. Procedia Engineering, 84, 12-22.
http://doi.org/10.1016/j.proeng.2014.10.405
Winfield, A. F., & Jirotka, M. (2017). The case for an ethical black box. In Gao, Y., Fallah, S., Jin, Y., Lekakou
C. (Eds.). Towards Autonomous Robot Systems, Lecture Notes in Computer Science, (Vol 10454, pp. 1-12).
Springer. http://doi.org/10.1007/978-3-319-64107-2_21
Woods, D. D. (2018). Decomposing Automation: Apparent Simplicity, Real Complexity. In Parasuraman, R. &
Mouloua, M. (Eds.), Automation and Human Performance: Theory and Applications, (pp. 3-18). Taylor and
Francis Group. https://doi.org/10.1201/9781315137957
Woods, D. D. (2016). The Risks of Autonomy: Doyle’s Catch. Journal of Cognitive Engineering and Decision
Making, 10(2), 131-133. http://doi.org/10.1177/1555343416653562
Woods, D. D., & Hollnagel, E. (2006). Joint Cognitive systems: Patterns in Cognitive Systems Engineering.
Taylor and Francis Group. http://doi.org/10.1201/9781420005684

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 38

Woods, D. D., & Sarter, B. N. (1998). Learning from Automation Surprises and “Going Sour” Accidents. In
Sarter, N. B., Amalberti, R. (Eds.). Cognitive Engineering in the Airline Domain. Taylor and Francis Group.
(pp. 327-354). https://doi.org/10.1201/b12462
Woodward, M., & Finn, C. (2016). Active One-shot Learning. Deep Reinforcement Learning Workshop, Spain.
https://arxiv.org/abs/1702.06559
World Intellectual Property Organization. (2019). WIP Technology Trends 2019: Artificial Intelligence.
https://www.wipo.int/edocs/pubdocs/en/wipo_pub_1055.pdf
Xu, X., Wickens, C. D., & Rantanen, E. M. (2007). Effects of conflict alerting system reliability and task
difficulty on pilots' conflict detection with cockpit display of traffic information. Ergonomics, 50(1), 112-130.
http://doi.org/10.1080/00140130601002658
Yeomans, G. (2014). Automation vehicles handing over control: opportunities and risks for insurance. Llyod’s.
https://www.lloyds.com/news-and-insight/risk-insight/library/technology/autonomous-vehicles
Zittrain, J. L., Barabas, C., Dinakar, K., Ito, J., & Virza, M. (2018). Interventions over Predictions: Reframing
the Ethical Debate for Actuarial Risk Assessment. Fairness, Accountability and Transparency in Machine
Learning. Proceedings of the 1st Conference on Fairness, Accountability and Transparency, USA.
https://dash.harvard.edu/bitstream/handle/1/34830702/ActualRiskAssessment_Feb_2018.pdf?sequence=1&is
Allowed=y

Authors

Dr Todd Pascoe, MOccMedHlth&SaF, PhD is a Business Improvement

Specialist in the Western Australian Mining Industry. Todd has worked in the
Western Australian Mining Industry for over 10 years in specialised roles
across Health and Safety, Mining Production and Business Improvement. Todd
has extensive operational experience in the deployment and day-to-day running
of an autonomous haulage system, including the development of risk
assessments, health and safety systems and operational efficiencies.

Dr Shirley McGough, RN, MHN, Dip App Sc, BSc (Nursing), PG(Nurs),
MNurs, PhD, Senior Fellow, Higher Education Academy. Shirley currently
works at Curtin University and teaches mental health, behavioural and
biosciences, conducts research and evidenced based practice. Shirley has
extensive clinical and research experience in mental health and wellbeing, with
substantial educational and leadership experience in the University sector.

Dr Janis Jansz, RN, RM., Dip. Tch, BSc. Grad. Dip. OHS, MPH, PhD, FSIA
is an Associate Professor in Occupational Health, Safety and Environmental
Health in the Western Australian School of Mines: Minerals, Energy and
Chemical Engineering at Curtin University in Western Australia and is a
Professor at the Xi’an University of Science and Technology, China. She also
works for the Healthforce Group. Janis is the Director of the World Safety
Organization National Office for Australia and Vice President of the
Occupational Health Society of Australia. She has been awarded Life
Membership of the Australian Institute of Health and Safety for many years of work improving,
teaching and conducting research to advance occupational safety and health practices and for taking a
leadership role the safety and health profession.

https://doi.org/10.5281/zenodo.6420932
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 39

World Safety Journal

A peer-reviewed journal,
published by the World Safety Organization

Journal Homepage:
https://worldsafety.org/wso-world-safety-journal/

Analysis of Australian Coronial Inquest and Non-Inquest findings of heavy vehicle

fatal crashes reveals shortcomings in investigations
Ivan Cikara1*, Geoff Dell2, Aldo Raineri1
1
Central Queensland University
2
Central Queensland University and VŠB-Technical University of Ostrava

KEYWORDS ABSTRACT
Heavy-Vehicle Transport Heavy vehicle crashes occur daily where drivers, occupants and other road users
Heavy-Vehicle Crashes are either killed or seriously injured. Investigations are conducted by regulatory
Coroner authorities of these crashes and reports are submitted to the Coroner to determine
Coronial Inquest the cause of death, make findings regarding the cause of the crash and then make
Coronial Non-Inquest
recommendations to improve safety and mitigate the risk of a crash occurring. The
Crash Investigations
Coroner plays a critical role in examining the cause of these crashes; however, this
research has identified there are a number of substandard investigative practices
where the investigations have not obtained the level of detail that would assist the
Coroner in making appropriate findings. This study has also identified that this
sub-optimal practice has been ongoing for a considerable period of time with a
number of Coroners expressing their concerns. This study has reviewed a total of
34 publicly available Coronial Inquest and Non-Inquest findings. The review has
identified there are inconsistencies in the quality of the investigations. A fatal crash
investigation properly conducted can be a valuable tool in identify why a crash has
occurred and greatly assist the Coroner the opportunity to make recommendations
from the findings to improve heavy vehicle transport safety.

1. INTRODUCTION

C
oroners plays a critical role in examining the underlying causes of the deaths and have the
opportunity to make recommendations from findings to improve safety (Brodie et al., 2010).
Coroners however rely on the information presented to them from the fatal crash investigation
report or from the information presented during an Inquest if one is held. The quality of that
investigation or the Inquest is dependent on the experiences, competencies and knowledge of the
investigator and the type of investigative methodology being used, if any (Cikara et al., 2020).

A fatal crash investigation, properly conducted, can be a valuable tool in identifying the underlying
causes and contributory factors of a crash, why a crash has occurred and suggesting recommendations
for future prevention (Brodie et al., 2010; Klockner & Toft, 2014; Dell, 2015; Dell, 2019). In most
instances the heavy vehicle fatal crash investigation is conducted on behalf of Coroners by the police.


Corresponding Author: [email protected]

https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 40

As stated by various Coroners, police may not have the knowledge, skills, competencies and resources
to understand and navigate the complexities of the heavy vehicle transport industry socio-technical
system and in some instances do not know what is needed to investigate a heavy vehicle crash to
identify the underlying causes (Productivity Commission, 2020).

The role of police is diverse, as well as complex, with competing priorities. In most instances police
investigate matters for prosecution purposes or for identifying liability (Selk & Benner, 2019; Doecke
et al., 2020; Productivity Commission, 2020) and their investigative training is not specific to the heavy
vehicle transport industry (Productivity Commission, 2020), with most investigators either learning
their skills on the job or bringing those skills acquired elsewhere with them which may not be
conducive to the regulatory setting they are working within.

2. CORONERS
A Coroner’s primary role is to establish the cause and circumstances of a sudden or unexplained death.
The role of a Coroner is fact finding, inquisitorial rather than fault finding, adversarial (Burns, 2014).
Deputy Coroner Loch of the Office of the State Coroner Queensland (2014) in a Finding of Inquest
(File 2012/334) stated ‘An Inquest is not a trial between opposing parties but an inquiry into the death.
The focus is on discovering what happened, not on ascribing guilt, attributing blame or apportioning
liability’ (p. 2). Chief Justice Neal Maclean described New Zealand’s coronial process as speaking
‘truth to power’ and the New Zealand Coroners website explains that a ‘Coroner speaks for the dead to
protect the living (Burns, 2014). The Western Australian State Coroner has stated that: ‘It is said that
the role of the Coroner’s Court is to speak for the dead and to protect the living. This two-fold role is a
vital component of a civil society’ (Fogliani, 2017, p. 5).

Information drawn from a Coronial Inquest is vital to facilitate risk identification and strengthen
countermeasures (Brodie et al., 2009). The significance of Coroners’ recommendations and comments
is their ability to advocate for change and potential wide-ranging impact (Brodie et al., 2010). In a
study to quantify Coroners’ recommendations and examine the nature of those recommendations
according to public health principles of injury causation and prevention, Bugeja et al. (2012) stated that
‘exploring the characteristics of recommendations generated from medicolegal death investigations is
an important step towards improving their contribution to injury prevention’ (p. 326).

Throughout Australia, Coroners’ courts are state based jurisdictions where it is a legislated requirement
that all deaths, including heavy vehicle fatal crashes must be reported to the Coroner in the State or
Territory in which that fatal crash occurs. Generally speaking, any unexpected, unnatural or violent
death must be reported. Internationally, within the Commonwealth, the requirement is not too
dissimilar. In fact, throughout the United Kingdom and countries of the British Commonwealth there is
much in common (Studdert & Cordner, 2010).

Coroners undertake Inquests based on the information gathered by police during the investigative
process. Studdert and Cordner (2010) identified that a Coroner’s role in establishing cause of deaths is
especially valuable for fatalities in which the fundamental cause is obscure, or first impressions are
misleading, which can be the case in heavy vehicle crashes. However, it was found that Coroners have
limited resources where accurately identifying causes of deaths requires prudent efforts and directions
towards investigations (Studdert & Cordner, 2010). Studdert and Cordner (2010) identified that
coronial investigations in Australia changed basic presumptions about how deaths occur only in a small
number of cases. This is concerning considering the findings of previous research that expressed
concerns at the lack of quality of investigations of heavy vehicle crash fatalities (Victorian Parliament
Law Reform Committee Review, 2006; Brodie et al., 2009; Brodie et al., 2010). This is made even

https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 41

more complicated where there has been a continuous effort to apportion blame for crashes on those
directly involved such as the driver of a vehicle and then taking legal action against them (Dell, 2015).

3. AIM
The aim of this research project was to analyse the findings of Coronial Inquests and Non-Inquests to
identify whether investigations of heavy vehicle fatal crashes by police provided sufficient evidence of
crash causation for Coroners to make informed decisions and recommendations to improve safety in
the heavy vehicle transport industry.

4. METHOD
A search was conducted of each Australian State and Territory-based Coroner’s office website.
Publicly available coronial reports were searched to identify those reports that included a crash
involving a heavy vehicle. The search was conducted for the years between 2005 and 2020. The reports
were either those arising from an Inquest as well as findings from a Non-Inquest. In total 34 reports
were selected. Each report was downloaded and read to ensure the report fell within the date range and
included a crash involving a heavy vehicle. The reports selected came from the Coroners Courts in the
states of New South Wales, Tasmanian, Victorian, South Australian, Queensland and Western
Australian.

The analysis identified the type of crash, i.e., truck vs car, the type of report, i.e., Inquest or Non-
Inquest, the findings and recommendations, the number of recommendations and types of
recommendations. The reports were further analysed to identify:

• If the crash was considered a workplace crash or not;

• The type of investigative methodology used by investigators;

• If comments were made by the Coroner with regards to the quality of the investigation and
information provided;

• Who was involved in the investigation process i.e., Police, Worksafe, National Heavy Vehicle
Regulator, Office of Transport Safety Investigations, Office National Rail Safety Regulator or
others?

5. CRASH TYPE
The search results identified 34 Coroners reports comprising of 13 x truck vs. car, 11 x truck rollover; 2
x truck vs. cyclist, 1 x lost load from truck vs. car, 1 x truck trailer vs. car, 1 x truck vs. tree, 1 x truck
vs. building, 1 x truck vs. motorcycle, 1 x truck vs. pedestrian, 1 x bus vs. pedestrian; and 1 x truck vs.
train.

6. CRASH TYPE AND WITH RECOMMENDATIONS

Recommendations arising from the coronial Inquests and Non-Inquest findings included:

1. Truck vs. train:

This was a workplace incident. The heavy vehicle driver was charged with criminal offences
however was found not guilty at trial. Two recommendations arose from the Non-Inquest.
Both recommendations related to the locomotive crashworthiness and structural integrity of
https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 42

the locomotive driver’s cabin. The Office of Transport Safety Regulator was involved in a
parallel investigation providing information to the Coroner. WorkSafe was not involved in the
investigation process.

2. A heavy articulated vehicle collided with a car:

Eight recommendations were submitted by the Coroner. These recommendations included
imposing speed restrictions for the articulated vehicle, instal internal speed limiters to be set at
60kph or at 80 kph if electronic stability control was fitted, driver must undertake practical and
theoretical assessments, amend the articulated code of practice to include guidance on
handling characteristics.

3. Truck vs. car:

The heavy vehicle rolled over rounding a bend, toppling over onto a car. Recommendations
included government to investigate stability control systems for heavy vehicles and for the
Queensland Department of Transport to appoint a coronial officer as well as mandate the
attendance of a transport inspector to the scene of fatal crashes involving heavy vehicles.

4. Truck vs. pedestrian:

This Inquest took into account five separate fatalities and recommended the Vehicle Standards
Australian Design Rules 42/04- General Safety Requirements 2005 be amended to require
front warning sensors be installed during manufacture of all cab over heavy vehicles.

5. Oversized escorts Inquests:

There were two Inquests that analysed three fatal crashes. In total 12 recommendations were
made. These included only granting permits for loads that are indivisible, not granting permits
for loads if they can me made smaller, night-time escorts to be limited to dual carriageways
and metropolitan areas, Police review the use of motorcycles as escort vehicles, oversized load
activity must be well managed, the National Heavy Vehicle Regulator regard the evidence of
the Inquest to develop regulations and guidelines for arrangements of oversized loads,
ensuring the wording for oversized loads are more effectively communicated, public
awareness campaigns ensuring public are aware of their obligations, traffic management plans
must be developed and distributed to all escort parties for the escort, procedures and permit
conditions to be more explicit when describing risks of transporting a wide load, using more
information signs for very wide loads, more attention be given to the spacing of the escorts
and the oversized load, must include this in the training for the Police officers escorting
oversized loads, lighting of oversized loads to be review.

6. Unstable trailer towed by truck colliding with a vehicle:

Recommendations included Government request training providers review content of training
and assessment for license holders, whether competence for loading and towing trailers should
be included in the training, government consider whether medium rigid licence holders ought
to be permitted to tow trailers behind medium rigid trucks, government to raise this issue with
other government agencies for consideration.

7. Car struck by heavy log falling from a truck:

Coroner recommended that WorkSafe review its industry standards and guidelines to address
aspects of the standards that were shown and conceded to be deficient during the Inquest.

https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 43

8. Truck colliding with a push cycle:

It was recommended that Government immediately prohibit cyclists from using the highway,
government engage with stakeholders to deal with safety concerns and provide an alternative
route, signage should be erected on onramps to warn cyclists that the roadway was dangerous.

9. Truck vs. car, double fatality:

It was recommended that government and Police maintain their awareness campaigns of the
risks of using mobile phones whilst driving and ongoing attention be given to detecting those
who breach the rules and consideration be given to the level of penalty applicable to such
offences.

10. Truck vs. tree:

It was recommended the shire conduct a risk assessment of the road with a view to placing
signage at the beginning of the road to forewarn road users of the road’s curves and
undulations and the dangers within.

11. Truck colliding with a push cyclist:

Seven recommendations were made. Some included the routine and mandatory electronic
recording of witness statements, government to regulate that vehicle stop one vehicle behind
the bike box which enables drivers to see the bike. Targeted and education program to alert
motorists of the risk of placing themselves immediately in front of a heavy vehicle, make such
actions an offence, conventional shaped heavy vehicle should be fitted with technologies to
warn drivers of obstacles and other road users in the driver’s blind spot, educate all motorists
about blind spots in front of a conventional heavy vehicle.

12. Bus hitting a pedestrian:

It was recommended that government urgently review the operations of the traffic and
pedestrian crossings at intersections.

13. Truck vs. car:

The Coroner recommended that Government proceed with the design and construction of a
channelised right turn at the intersection with priority.

14. Truck vs. tree:

It was recommended that the government consider enhanced education and awareness to
ensure employees performing inspection of heavy vehicle are of the risks involved in not
having clean components to inspect, and that government review the current memorandum of
understanding between Police and Worksafe that includes a process for the notification of a
heavy vehicle incident to the national heavy vehicle regulator.

15. i) Truck vs. car & ii) truck vs. pedestrian:

One Inquest was held for two separate fatalities occurring at the same location. 17
recommendations were made (South Australian Government 2015) these included:

Increase penalties, including imprisonment, be enforced on drivers who do not use a safety
ramp when their truck is out of control, a truck driver automatically be charged with

https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 44

dangerous driving should they exceed the speed limit of 60kph on the down stretch of the
freeway, owners or companies employing drivers compensate victims and their families for
crashes involving heavy vehicles, truck drivers get compulsory training about driving downhill
and using arrester beds.

Trainee drivers should undergo specific tuition in relation to the required manner of driving on
the descent of the South Eastern Freeway, a truck driver must be supervised when driving
down the South Eastern Freeway for the first time and the accompanying driver must have
demonstrated experience and competence on that down-track, no heavy vehicle licence of any
kind should be issued to any person in Australia unless they have demonstrated competence in
the safe negotiation of the decent on the South Eastern Freeway.

Speed limit for heavy vehicles be reduced to 40km/h on the down track, improved signage,
including informing drivers that heavy penalties apply if trucks and buses do not use low gear,
promote arrester bed use, roadworthiness and maintenance be brought within the chain of
responsibility regime with the Heavy Vehicle National Law (South Australia) Act 2013 and be
undertaken nationally, all heavy vehicles be subjected to a periodic and frequent inspection
regime, investigating the capability of technology to detect the speed of a heavy vehicle and
tell a driver that their speed is excessive and they needed to use a safety ramp.

Consideration given to the creation of an area situated between the Heysen Tunnels and the
second arrester bed to be used for consideration be given to the creation of an area situated
between the Heysen Tunnels and the second arrester bed to be used for the mandatory
stopping of all heavy vehicles with a further requirement that if the vehicle is incapable of
stopping at that area, the driver must use the second safety ramp (Marcus, 2015).

However, in this Inquest the Government of the day stated it would only support 13 of the 17
recommendations (MacLennan, 2015).

16. Truck rollover:

No recommendations were made however a number of points were raised. Company did not
do any act or omission to contribute to the crash. Driver not wearing seatbelt. Delays in
completing investigation, investigation not completed to a required standard. Workplace
incident, WorkSafe not involved.

7. FINDINGS
Over the 15-year period this study identified a total of 34 publicly available Coronial Inquest and Non-
Inquest reports. In those 34 reports a total of 60 recommendations were made with five Inquests
providing 43 recommendations and the remaining 29 Inquest and Non-Inquest findings providing 17
recommendations. This study supports the findings of a previous study conducted by Brodie et al.
(2010) in which it was identified that Coroners recommendations from fatal crashes were infrequent.
The Inquests where multiple recommendations were made came about where Coronial Inquests were
combined and heard as one because the fatalities occurred at the same location or were similar types.
i.e., fatal crash occurring during the escort of an oversized load.

There was one Inquest that investigated two incidents because the fatalities that occurred were the
result of the same type of incident, this being a collision occurring as a result of escorting oversized
loads. From these two fatalities there were seven recommendations. There was a second Inquest that
also involved an oversized escort from which there were five recommendations.

https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 45

There was one Inquest that had a total of eight recommendations, one Inquest had six
recommendations. Twenty of the Coroners’ reports focused on the heavy vehicle driver behaviours. In
seven of the reports, it was noted the driver was criminally charged however the driver was found not
guilty in four of the crashes. In another crash the driver pleaded guilty, despite there being evidence to
suggest other contributing factors. In another crash the driver was convicted of the offences and in
another crash the charges were downgraded to lessor charges to which the driver pleaded guilty.

Thirty-three of the crashes involved a workplace incident as the heavy vehicle driver was working at
that time. Only one of the 33 crashes that were workplace incidents was investigated by WorkSafe.

The types and focus of recommendations made were varied. Most were directed towards the
management of driver behaviour such as imposing speed restrictions, driver training, installing speed
limiters in heavy vehicles, better risk management of oversized escorts, employees performing
inspection of vehicles and increased penalties for drivers who do not travel at restricted speeds. There
were recommendations that included increased penalties for breaching the laws, installing in-vehicle
technologies such as stability control and front warning sensors and other recommendations were
aimed at improving the manner in which investigations are conducted. For example, in one Inquest it
was recommended that there be routine and mandatory electronic recording of interviews, another
recommended the mandatory attendance of a transport inspector at the scene of a fatal crash and in
another that government review its current memorandum of understanding between police and
WorkSafe. Support, by way of applying a consistent investigative methodology as well as providing
training to investigators on the heavy vehicle transport system, is needed. This will enable the
collection of evidence and information to identify and analyse the underlying causes for targeted
prevention. As stated by Brodie et al. (2009) ‘Increased recognition of heavy vehicle safety beyond
individual driver responsibility is needed’ (p.563). In support Newnam et al. (2017) identified that a
comprehensive investigative process is likely to identify interventions that have contributed to a crash
all the way up to the level of regulatory bodies and Government agencies

More importantly, in 22 of the Coronial reports, it was identified that the heavy vehicle driver
behaviour, actions or omissions had in some part contributed to the crash. For example, in one Inquest
it was identified the driver had no experience driving a heavy vehicle and, in another Inquest, it was
found the driver was significantly fatigued as a result of being under too much pressure to meet a
deadline to deliver their load. However, the Inquest did not delve more deeply into the safety
management system to identify 1. If the driver’s employer had checked the drivers were competent and
experienced to drive a heavy vehicle 2. whether the scheduling and risk management processes
permitted a driver to drive a heavy vehicle to meet unrealistic deadlines causing them to become
fatigued.

In addition, the role and importance of systemic investigations was recognised. In an Inquest where
separate investigations had been conducted by police and the Queensland Office of Transport Safety
Investigations, the Coroner noted that police were statutorily prohibited from relying on the
investigation reports from another agency for prosecution purposes and that police had to
independently gather and analyse the evidence using their own expertise and resources. The Coroner
went on to further state that the other agency involved in the investigation was focused on identifying
contributory factors for the purpose of identifying opportunities to improve safety and health
(Queensland Coroners Court 2008/392 & 2008/393, p. 4).

https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 46

8. DISCUSSION
8.1 Quality of crash investigations

A review of the Inquest and Non-Inquest findings suggests that systemic investigations are not being
undertake. This is supported by comments made by Coroners. This study found there were a number of
comments made by Coroners regarding the quality of crash investigations. This has been consistent
since 2001 where Coroners around Australia have made similar findings and expressed their concerns.
For example, in a Victorian Coronial Inquest (1114/2001), the Coroner recommended that police
consider developing a basic investigation standard for fatal and serious crashes. This recommendation
came about due to a number of serious deficiencies identified by the Coroner of an investigation
conducted by the police of a fatal crash. Additionally, the Victorian Parliament Law Reform
Committee Review (2006) conducted a review of the Victorian Coroners Act 1985 and reported that
the limitations of Coroner’s investigations related to transport collision had been raised in a number of
forums over the previous past five years. The review identified that:

• Coroners did not have legal control over the death investigation process nor any power to direct
the investigation, and

• There was an absence of comprehensive guidelines for Coroners conducting death

investigations as well as widespread discrepancy between the standard of Coroners’
investigations in Melbourne and those in rural areas.

In a Coronial Inquest regarding a vehicle colliding with an escorted oversized load, killing the car
driver (Queensland Coroners Court 2199/05(0)), the family of the deceased were dissatisfied with the
evidence submitted to the Coroner by the police. The family of the deceased had taken it upon
themselves to assemble a substantial body of material regarding the conditions and events prior to the
collision and provided that information to the Coroner. That material was subsequently forwarded by
the Coroner to the police for further investigation who were then able to obtain further evidence for the
Coroner. Although not criticising the investigation, the Coroner found (Queensland Coroners Court,
2199/05(0), p. 28) that a Coronial Inquest is better informed when evidence is available from all
perspectives.

In another Findings of Inquest, (Queensland 2011/1631 & 2010/4091) a police officer riding a
motorcycle and escorting an oversized load was struck and killed by a heavy vehicle travelling in the
opposite direction. The heavy vehicle driver was charged with driving without due care and attention,
however, after a three-day trial, the heavy vehicle driver was found not guilty (p. 11). In the subsequent
Inquest, the deceased’s wife made a submission to the Coroner to consider flaws in the way the
investigation was undertaken and managed (p. 12). During the Coronial Inquest the Coroner identified
a number of ‘sub-optimal practices’ (p.12) that included:

• Conversations with witness at the crash scene were not tape recorded.
• Detailed accounts were not taken from any witnesses on the morning of the crash.
• Police did not have witnesses adopt any notes taken of the conversations.
• Key witnesses were allowed to confer at the scene and then leave on the understanding they
could attend to a Police Station at a time and place convenient to them to provide a statement.
• The defendant was spoken to by four officers on the day of the crash, but none took detailed
statement or questioned the defendant closely as should have been in such circumstances.

https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 47

• One officer was notionally put in charge, but other officers gave directions about aspects of the
investigation without consulting with the person in charge and did not ensure the person in
charge was advised of the outcomes.
• Other officers who were tasked with various inquiries or to obtain statements were inadequately
briefed regarding the issues under investigation as well as the versions provided by other
witnesses meaning they could contribute little to the investigation.
• The expert investigators from the Fatal Crash Unit did not arrive until two days after the crash.
• Significant delays occurred during the investigation as the supervisors allowed significant
delays to occur.
The Coroner noted that ‘Whilst these criticisms have some force, as a result of reviewing the
investigations of numerous traffic fatalities I am aware that similar practices are common’ (p. 13).

The Coroner went on to further identify a number of key points in relation to the quality of the
investigation. These included the investigating police officer worked in a very busy station, the officer
had very limited crash investigation training and experience, there was no specialist support, experts
undertook discreet parallel investigations or reviews, there was limited coordination of the number of
investigations being conducted at the same time. The Coroner noted that:
‘(name redacted) was the officer in charge of a very busy division with limited crash
investigation training and experience. Nonetheless, in the normal course he would
investigate fatal road crashes in that division, there being at the time no specialist FCU in
central region. However, in this case he was left to undertake a criminal and coronial
investigation of a particularly sensitive matter, while other experts undertook discreet
investigations or reviews of aspects of the incident. There was limited coordination of the
four inquiries and less supervision of (name redacted).
In conclusion, the Coroner stated that ‘[t]he outcome was less than perfect, but probably no worse than
many investigations regularly undertaken throughout the state’ (p.13).

These comments raise a number of concerns at the lack of quality of investigations that are conducted
of fatal crashes. The Coroner’s comments suggest that sub-optimal practices in investigations are
common place and the quality of heavy vehicle fatal crash investigations are not to a suitable standard.

The Coroner’s comments were consistent with the findings from research conducted by the authors of
this paper where a number of participants were interviewed to identify why heavy vehicle drivers were
blamed for crashes. The participants in the interviews, many being either police officers or former
police officers, stated that police officers, themselves included, were required to conduct heavy vehicle
fatal crash investigations whilst working under strenuous circumstances, without support, in busy
police stations with limited or no training or crash investigation experiences. This may, to some extent,
explain the sub-standard quality of investigations being conducted resulting in deficient investigation
reports being submitted to Coroners.

Research by Brodie et al. (2010) found there was no systematic approach to Coroners’ investigations of
heavy vehicle fatalities. The research identified that police crash expert investigators are not involved
in the investigation of all fatal heavy vehicle crashes, especially those that are single heavy vehicle
crashes. Generally, these investigations are conducted by general duties police often with less
experience and expertise that those within the specialised crash investigation squads.

In another Non-Inquest of a single heavy vehicle rollover fatal crash killing the driver (Magistrates
Court of Tasmania 2020), the Coroner found the police officer allocated to the investigation was not a

https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 48

qualified investigator and was attached to a police station with a very heavy workload. The Coroner
concluded that it was understandable the police officer was unable to complete the investigation to a
required standard and in a timely manner (Magistrates Court of Tasmania 2020, p. 7). These comments
are consistent with the research findings by the authors of this paper who conducted a number of semi-
structured interviews with a participant who had investigation experiences. The participants stated that
when working in busy police stations, particularly those police stations in country areas, they did not
have the support or resources to properly complete a heavy vehicle fatal crash investigation, nor did
they have the training, competencies or experiences to conduct those investigations.

In an Inquest finding of a fatality caused by an incorrectly loaded trailer causing it to lose balance
leading to a crash, the Coroner became aware of a number of systemic failings identified during the
investigation. The Coroner suggested that ‘where systemic flaws in the system are disclosed, they
should be addressed with urgency and careful attention’ (Coroner’s report 2013/298913, p. 11). Later
in the report, the Coroner stated in relation to the circumstance surrounding the death that:
Coroners have no professional expertise in relation to these questions. I have therefore
framed the recommendations not in rigidly prescriptive terms but putting the serious
issues (name redacted) death raises before the experts in the field for them to consider and
resolve (p. 13).
However, the substandard quality of investigations is not just specific to police. In another example of
an Inquest, the Coroner expressed concerns regarding an improperly conducted investigation by
another government agency. The Coroner commented on the substandard quality of a tape-recorded
interview, noting that:
The tape recording of the interview reveals a rather shambolic, rambling and unstructured
interchange. The panel chair stated at the Inquest he had taken some comfort from
knowing that a coronial investigation was to follow. However, subsequent events have
revealed this to be illusory. While there is no doubt the panel members had high level
content knowledge, it might be desirable in future to include a person with forensic
expertise (Queensland Coroners Court 2009, p. 3).
The research suggests there are sub-standard and sub-optimal investigation practices that have been
evident for a considerable period of time. Driscoll (2003) identified that limited investigative resources
and a lack of standard investigation approaches were likely to impede the information and investigative
process. Bohensky et al. (2005) identified that if the process for investigating deaths were standardised,
these deaths would be more easily and efficiently investigated. Bugeja et al. (2007) concluded there
were recognised limitations associated with the coronial investigations of heavy vehicle crashes and the
lack of an existing systematic approach affecting investigation outcomes.

A review conducted by the Government of Western Australia Department of Treasury (2018) to

improve the efficacy of coronial investigations recommended a number of efficiency and effectiveness
reforms, three of which focused the lens on the quality and output of investigations. These
recommendations included enhancing the powers of Coroners to obtain information outside of an
Inquest, improving consistency of regional investigations and updating guidelines to police and
modifying performance measures reflecting the integrated components of coronial investigations. The
recommendations highlight the need for police to conduct consistent investigations.

This is not new as previous studies by Brodie et al. (2009) posited that opportunities remain through
learnings from fatal crash investigations and the systematic analysis of contributing factors. In that
study it was concluded that a standardised approach to heavy vehicle fatal crash investigations will
support Coroners and research efforts to increase knowledge. In a subsequent study by Brodie et al.
(2010) into recommendations following fatal heavy vehicle crash investigations, it was identified that
https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 49

investigation standards were lacking. This included a need for a minimum crash investigation standard,
improved investigative practices and increased resources for data collection to enable proactive
measures by investigative authorities (p. 140).

9. CONCLUSION
The analysis of the coronial Inquests and Non-Inquest findings identified that crash investigations
differed in quality and complexity. This study identified there have been historical concerns raised by
Coroners over the substandard quality of investigations arising from reviews and Coroners’ findings.
This study has revealed a number of comments from Coroners who have continued to voice their
concerns of the quality of investigations and that it suggested that sub optimal practices are
commonplace. The investigations were all conducted by the Police for the Coroner from differing
states. The thematic analysis identified there were differing quality of investigation, some of which
were criticised by the Coroner.

A review of the coronial Inquests and Non-Inquest findings revealed that there did not appear to be a
standardised approach to the investigations. There were varying degrees of quality and detail as
evidenced by Coroners’ remarks and the investigations did not identify contributory factors across the
levels of the socio-technical system. Most factors identified in the coronial Inquests would more than
likely be limited in scope because the investigations did not appear to adopt a systematic approach. It is
also highly likely that other systems factors at the upper levels of the socio-technical system were
present however not identified nor investigated due to the investigator’s lack of understanding and
knowledge of effective investigation practices.

REFERENCES
Bohensky, M., Ibrahim, J.E., O’Brien, A.L., Emmett, S.L., Newman, E., Charles, A., & Young, C. (2005).
World without borders. Integrating clinical perspectives into coronial jurisdiction in Victoria, Australia.
Journal of Medicine and Law, 25(1), 13-29.
Brodie, L., Bugeja, L., & Ibrahim, J.E. (2009). Heavy vehicle driver fatalities: Learnings from fatal road crash
investigations in Victoria. Journal of Accident Analysis and Prevention, 41, 557-564. DOI:
10.1016/j.aap.2009.02.005.
Brodie, L., Bugeja, L., & Ibrahim, J.E. (2010). Coroner’s recommendations following fatal heavy vehicle crash
investigations. Australian and New Zealand Journal of Public Health, 34(2), 136-141. DOI: 10.1111/j.1753-
6405.2010.00497.x.
Bugeja, L., Symmons, M., Brodie, L., Osborne, N., & Ibrahim, J.E. (2007). Development of a specialist
investigation standard for heavy vehicle fatal collisions in Victoria. In P. Schofield (Ed.): Conference
Proceedings of the 2007 Australasian Road Safety Research, Policing and Education Conference. Melbourne
Victoria Australia: The Meeting Planners, 1-10.
Bugeja, L., Ibrahim, J.E., Ozanne-Smith, J., Brodie, L.R., & McClure, R.J. (2012). Application of a public health
framework to examine the characteristics of coroners’ recommendations for injury prevention. Journal of
Injury Prevention, 18, 326-333. DOI: http://dx.doi.org/10.1136/injuryprev-2011-040146
Burns, B. (2014). Lessons learned from the Coroner’s court. Kai Tiaki Nursing New Zealand, 20(8), 21-23.
Cikara, I., Dell, G., Toft, Y., Dell, S., & Skrizovska, M. (2020). The need for enhanced heavy vehicle crash
investigations: Beyond ‘blame the driver’. Transactions of the VSB - Technical University of Ostrava, Safety
Engineering Series, 15(1), 60-72, (ISSN: 1805-3238). DOI: 10.35182/tses-2020-0007.
Dell, G. (2015). Foreword, in Viner D, Occupational risk control: Predicting and preventing the unwanted.
Gower Publishing, Farnham.
Dell, G. (2019). The lessons from 15 years of airline corporate governance roles (and 40 years of air safety
investigations). Presentation at the Australia and New Zealand Societies of Air Safety Investigation,
Wellington, New Zealand.
https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 50

Doecke, S., Thompson, J., & Stokes, C. (2020). How do we prevent and mitigate crashes? Evidence from
Australian at-scene in-depth crash investigations. Journal of Safety Research, 1(2), 35- 43.
Driscoll, T. (2003). Traumatic work-related deaths and Australian Coroners. A thesis for preventions. The
Journal of the Australasian Coroners Society Inc, Issue 2.
Fogliani, R. (2017). Office of the state Coroner for Western Australia. Annual Report 2016-2017.
https://www.coronerscourt.wa.gov.au/_files/Annual_Report_2016_17.pdf
Government of Western Australia, Department of Treasury. (2018). Improving the efficiency of coronial
investigations, 90 day regulatory mapping and reform project series report no 4.
Heavy Vehicle National Law (South Australia) Act 2012.
Klockner, K., & Toft, Y. (2014). Accident modelling using social network analysis for complex socio technical
systems . Australian Safety System Conference (ASSC 2014), Melbourne, Victoria, Australia.
Marcus, C. (2015). Coroner recommends national training for truck drivers on the South Eastern Freeway into
Adelaide. ABC News [online]. https://www.abc.net.au/news/2015-01-12/Coroner recommends national
training for truck drivers on the South Eastern Freeway/6012304. [Accessed 23 June 2021].
MacLennan, L. (2015). Truck drivers could be jailed if low gear not used on South Eastern Freeway decent in
Adelaide Hills. Truck drivers who fail to engage low gear when descending the South Eastern Freeway in the
Adelaide Hills could be jailed under legislation the South Australian Government is drafting, Sydney.
Magistrates Court of Tasmanian. (28 October 2020). Record of investigation into death (without Inquest), James
William Espie.
Productivity Commission. (2020). National Transport Regulatory Reform, Inquiry report: Overview and
recommendations. Australian Government, report no. 94, Canberra. ISSN 1447-1329-1447-1337.
Newnam, S., Goode, N., Salmon, P., & Stevenson, M. (2017). ‘Reforming the road freight transportation system
using systems thinking: An investigation of coronial Inquests in Australia. Journal of Accident Analysis &
Prevention, 101, 28-36. DOI: http://dx.doi.org.exyproxy.lib.monash.edu.au/10.1016/j.aap.2017.01.016
Queensland Courts, Office of the State Coroner. (22 December 2008). Findings of Inquest, Innisfail Coroners
Court, file number 2199/05(0).
Queensland Court, Office of the State Coroner. (23 December 2009). Findings of Inquest, Brisbane Coroners
Court, file number COR868-870/06.
Queensland Courts, Office of the State Coroner. (15 March 2013). Findings of Inquest, Brisbane Coroners Court,
file number 2011/1631 & 2010/4091.
Queensland Courts, Office of the State Coroner (5 June 2014). Findings of Inquest, Toowoomba Coroners Court,
file number 2012/334.
Queensland Courts, Office of the State Coroner. (27 March 2018). Findings of Inquest, Toowoomba Coroners
Court, file number 2014/3137.
Queensland Coroners Court, Office of the State Coroner, Findings of Investigation: Non-Inquest findings into
the deaths of Michael Smithers and Richard Wetherell, 2008/392 & 2008/393, Cairns, Australia.
Selk, S., & Benner, L. (2019). Accident investigation challenges for process safety practitioners. Journal of
Process Safety Progress, 28(4), 1-9.
South Australian Government. (2015). Response to the deputy State Coroner’s recommendations of January
2015, following the death of Mr James William Venning on the South Eastern Freeway, September 2015.
State Coroners Court. (13 June 2016). New South Wales, Findings from Inquest, 2013/298913.
Studdert, D.M., & Cordner, S.M. (2010). Impact of coronial investigations on manner and cause of death
determinations in Australia, 2000 – 2007. Medical Journal of Australia, 192(8), 444-447.
DOI:10.5694/j.1326-5377.2010.tb03582.x
The Coroners Act 1985 (Victoria)
Vehicle Standards Australian Design Rules 42/04- General Safety Requirements 2005
Victorian Coroners Court. (2001). Case number 1114/2001.
Victorian Parliament Law Reform Committee. (2006). Inquiry into the review of the Coroners Act 1985.
Parliamentary Paper No. 229 of Session 2003-06. ISBN-0-9757984-1-3, Melbourne Victoria.
https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 51

Authors
Ivan Cikara is a PhD Candidate at Central Queensland University completing
his PhD on heavy vehicle fatal crash investigations titled ‘Systemic
investigations are needed to improve safety in the heavy vehicle transport
industry’. Ivan has a Master’s degree in Occupational Health and Safety,
Masters in Business Administration as well as a number of Advanced
Diploma’s and Diplomas. Ivan has over 30 years’ experience working in both
the Government and private sectors. This includes working in Policing where
the main focus of his duties was on fatal crash investigations, before
progressing to the Homicide Squad as a Senior Detective. Ivan’s career
continued into investigations management in the medical and health industry as
well as safety and investigations management in the transport, maritime, rail, mining and oil and gas
sectors, holding a number of senior executive roles. Currently Ivan is the Group Safety Manager for a
national organisation. Previously, Ivan worked as the National General Manager of Safety for a
national transport company as well as being the Corporate Manager of Safety Investigations for a
Government transport provider.

Dr. Geoff Dell is a Professor at Ostrava University and Adjunct Professor at

Central Queensland University. Dr Dell was also the former Head of Safety
Science Courses. He is a career Safety Scientist with 40 years of senior safety
management roles in aviation, rail and other industries. Dr Dell was National
President of the Safety Institute of Australia for 9 years and the Independent
Safety Adviser and Member of the Board of Directors of Cobham Aviation
Services, the third largest airline in Australia, and was the independent
Chairman of the Board Safety Sub-committee of Nauru Airlines for 2 years. In
2010, he joined CQUniversity to develop and implement the world first
tertiary education courses in accident investigation including the unique
Bachelor of Accident Forensics and the Master of Accident Investigation with majors in air safety
investigation, rail safety investigation, road safety investigation, occupational safety investigation, risk
engineering and human factors engineering. This work included the development of the world’s first
genuine multi-modal forensic investigation teaching laboratory (Crash Lab). He has over 40 papers
published in peer reviewed and refereed journals, has presented over 100 invited papers at safety
conferences globally and is sought after by the media in Australia and overseas as a commentator on
aviation and other crashes, including MH370, MH17, German Wings 9525, Air ASIA 8501 and more
recently the Lion Air and Egypt Air B737 Max crashes

Dr. Aldo Raineri is currently the Head of Courses – Safety Science, CQ

University Brisbane. In December 2011 Dr Raineri retired as the Director of
Work Health & Safety Policy & Legislation, after having spent over 20 years in
various management and safety roles with the Queensland Government. During
that time, Dr Raineri was involved in every major strategic and operational
government safety initiative, including the national development of harmonised
work health and safety laws and their implementation in Queensland. Dr
Raineri has also been a sessional lecturer, tutor, course developer, educational
administrator and project supervisor in the TAFE system and at a number of
Universities in Australia for the past 30 years.

https://doi.org/10.5281/zenodo.6420953
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 52

World Safety Journal

A peer-reviewed journal,
published by the World Safety Organization

Journal Homepage:
https://worldsafety.org/wso-world-safety-journal/

Fast-track Industry Safety Culture and Reduction of Incidents/Losses

Harbans Lal
A Retired Professor of Psychology, SNDT Women's University, India

KEYWORDS ABSTRACT
Behaviour The term "safe culture" refers to people's safe habits. Safety culture is on the rise in
Risk today's industry. A safety culture must be developed to prevent workplace
Safety mistakes and accidents. Safety culture philosophies vary by industry in terms of
Culture practices. The goal of this article is to identify key issues in establishing a
Qualitative
behavioral safety-friendly culture in business, as well as potential solutions. Ten
themes that address critical issues, such as fundamental questions about long-term
safety cultures, reactive safety cultures, collective voice and leadership for at-risk
behaviors, religion, spirituality, and festivities for site safety, and safety
implementation with empathy for others, are presented.

1. INTRODUCTION

I n a favorable safety culture, workers choose safety in everything they do, even if it goes beyond
industry rules. In a bad safety culture, safety is an afterthought, or a delayed step (ISHN, 2021). A
supportive safety culture helps to ensure everyone’s safety.

In recent years, the concept of cultivating a positive safety culture has gained much attention
(Williams, 2021). The safety culture of an organization is attributed to it in a number of ways. A
sequence of small acts of kindness and adjustments in safety habits can have a big influence in the long
run (Pettinger, 2020).

Safety cultures can persist for a long time when they are integrated into a strong company culture
(Paoletta, 2020). The safety culture improves dramatically when companies involve all employees in
the process of risk control through observation and spot-correction. In terms of long-term safety culture
management, the employer’s aphorism is a high-risk proposition for long-term economic survival
(Kaila, 2021). The company's safety culture both supports and propels it forward.

The organizational behavior culture is to blame for thousands of deaths and injuries. To remove at-risk
behaviors, plant employees must appreciate safe behaviors, which they develop on a daily basis as
observers of the safety culture. Every plant requires immediate changes to its safety culture, which can
only be carried out by people. The majority of safety professionals are trained in safety systems rather
than safety culture development, which is a critical issue. This article offers recommendations in this


Corresponding Author: [email protected]
https://doi.org/10.5281/zenodo.6420967
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 53

respect (Kaila, 2021). The goal of safety culture is to serve humanity without causing mishaps or
accidents; that is the essence of safety culture!

Zero-risk management necessitates a variety of behavioral patterns from top to bottom, as well as
among colleagues. The road to get there, on the other hand, is both entertaining and enlightening, as
well as heartbreaking. In this regard, there are numerous organizational antecedents that must be
considered. Employees perceive change within an organizational safety culture as a result of
involvement through processes of observations and spot-corrections. Kaila (2021): When this
intervention is done on a regular basis and employees like it, the business's values become part of the
way they do business.

2. BEHAVIORAL RISK MANAGEMENT

IN THE CONTEXT OF SAFETY CULTURE

The fundamental fault is that there are no safety regulations in place. Systematized, common sense
practice has become a standard in the industry. There are still a lot of questions about behavioral risk
management for safety culture adoption that need to be addressed, as one goes through the safety steps
(defining, systems, culture, and management). Below are some alternative answers.

Theme 1: Fundamental questions about creating a long-term supportive safety culture

In India, the size of the safety net for employees is determined by the size of the employer's pocket.
The amount of profit made by the organization determines whether employees need a safety system or
not. Care is a distant second; money comes first. Consequently, at-risk employees' lives are at risk, and
employers' businesses are at risk too, as a result of at-risk behavior on the job. To ensure the long-term
viability of businesses, a culture that prioritizes employee safety must be adopted. Due to various
variables at sites, it can be difficult for even leaders to persuade some employees that HSE infractions
are not acceptable. It must be recognized that high-risk behaviors in the workplace are regarded as
dangerous to citizens in civilized societies and can be debilitating, harmful, or even lethal. To develop a
safety culture, spot-correction is a function for everyone. Businesses appear to have a grim future if
they do not build a long-term, supportive safety culture at their sites.

By all means, a safety culture is "a set of common practices and behaviors that are used by everyone in
an organization to try to control the most important risks associated with its work" (Velas, 2021).

Theme 2: A reactive safety culture

A reactive culture indicates that safety mechanisms are patched together and developed in response to
negative incidents and legal obligations (Halligan et al., 2013). Why are industry professionals so
adamant about maintaining a reactive safety culture? "It is their comfort zone, and they don't want to
feel discomfort," says Pavan Rao of Sembcorp Nellore. According to KK Sharma of DCM Shriram,
those who are reactive at sites are those that are self-centered and need to demonstrate their strength,
while others require attention, guidance, or instructions for important tasks to be completed, owing to
their upbringing or culture. Safety is an afterthought in a weak or reactive safety culture, whereas in a
positive safety culture, safety is pushed beyond industry requirements (ISHN, 2021). At the NTPC
Safety Academy, A. K. Dang says that behavior-based safety plays a big part in accident prevention
and intervening behavioral science. This is when site management, including contractors, requires
people to be proactive rather than reactive in identifying at-risk behaviors in order to meet the zero-
accident goal.

https://doi.org/10.5281/zenodo.6420967
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 54

Theme 3: At-risk behaviors, the collective voice of observers and leadership

The power of a group's collective voice outweighs the threat of any type of punishment. At-risk
behaviors serve as a warning signal, right before an incident occurs. The most crucial aspect of
observer training is to comprehend the daily observations in terms of the company's safety culture.
Building a safety culture requires both personal and organizational safety practices. Urbach and Fay
(2020) found that the quality of the leader-member exchange was a key predictor of whether leaders
would back employees' ideas for positive change.

Leaders rarely conduct observation rounds on their own; instead, they present observations made by
their observers to the lower levels. Incidents are also triggered by hotspots of at-risk behaviors, and
while spot-correction of at-risk behaviors is done, behavior must be altered through regular risk-based
dialogues (RBC). This is critical feedback that must be addressed in order to maintain a safe work
environment. As a result, long-term safety culture management requires a strong leader who can guide
and direct everyone from the top down.

Theme 4: Religion, spirituality, and festivities are important forms of multicultural diversity
for site safety

People who feel spiritually protected are more likely to care for others at work, thus reducing
workplace dangers. When people respect one another's spiritual principles, they form strong bonds that
help to foster a sense of fraternal safety (Keenan, 2017). Spiritual beliefs and HSE beliefs are
intertwined because they are based on human experience (Kaila, 2021). Allowing for variances in
religious views and spirituality in the workplace promotes the creation of a psychologically healthy
working environment (SHRM, 2020).

Safety culture is a journey, not a goal. By all means, a strong safety culture within a business has been
found to cut down on the number of workplace injuries and accidents.

Theme 5: The execution of safety with consideration for others

HSE systems are maintained by management. They must ensure that the behaviors associated with
these systems are triggered. For the HSE's cultural transformation to succeed, industry-wide action is
needed. Implementing behavioral safety is an extension of care to culture from the workplace to every
place (Kaila, 2021). Spot-appreciation of safe actions and spot-correction of risky behaviors are both
important to improving safety culture.

For corporations, safety is a priority; for contractors, safety is a must. The consequences of safety
implementation differ dramatically and are challenging to achieve. By all means, excellence in safety
culture leads to business sustainability. To have a truly comprehensive safety program, managers need to
include psychological safety (Barnes, 2021).

Theme 6: Companies that do not enable their employees to carry out safety implementation

Building a safety culture can help businesses succeed in their safety processes (Vos, 2021). If one is
truly interested in creating a strong safety culture, then s(he) must listen to every observer who instills
in each employee a desire to behave safely when they are behaving unsafely (Kaila, 2021).

https://doi.org/10.5281/zenodo.6420967
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 55

By all means, a low budget for safety is equivalent to a large budget for mishaps. Mentors who are
talking about safety culture engagement should think about real-world examples of safety culture
changes that have been shown to work (ISHN, 2021).

In order to effectively implement behavioral safety at workplaces, leaders must incorporate a safety
culture into their personal lives.

Theme 7: The safety professionals' competencies gap

Competency development is a process that lasts a lifetime. There are numerous aspects that influence
an individual's ability to achieve competency: age, gender, education, geography, local influences,
family background, and environment are just a few examples. When establishing competency, these
criteria are never taken into account as a whole. Because of the lack of consistency in judging
competency, there is always a gap that is never filled.

By all means, the world is shifting away from industrial safety toward a new perspective on safety.
Human Factors, Human Performance, Psychological Safety, and so forth, are all new concepts. These
HSE viewpoints are not taught to safety experts. The job description of HSE experts seems to be
changing around the world, which means that they need new skills (Burdick, 2019).

Theme 8: Major obstacles to taking health and safety decisions

When firms abuse safety, safety specialists become frustrated; but they should not be made scapegoats
(MySafetySign, 2021). By all means, maintaining the highest level of long-term safety culture is a
never-ending journey for management and workers on the job. A positive outcome in terms of safety
necessitates honest practice by all parties involved. It will be incredibly difficult for business
executives to save their companies and their employees if they do not prioritize safety over commercial
operations (Kaila, 2021).

Some so-called BBS (Big Brother/Big Sister) specialists are attempting to teach behavioral safety
culture in a muddled manner due to a lack of understanding of the BBS approach to behavioral science.
This sends the wrong impression to businesses about what BBS is and is not, as well as how to
scientifically adopt it. The BBS journey is beneficial to the company's financial security since it allows
management to say goodbye to incident costs, compensation, and litigation. A safe workplace is one
where everyone gets home without being hurt. (Kaila, 2021).

Theme 9: Top leaders' spot-implementation of the BBS approach

The crucial question is how quickly a safety culture can be instilled in order to save everyone at work.
To reinforce spot-implementation of the BBS strategy for quick results and a long-term safety culture
at sites, it requires highly brave, fast-track leadership.

By all means, if the safety culture is accurate and faultless, the company will save a lot of time. Safety
procedures will be driven from the top down in the perfect safety culture. Practices for safety will filter
down to the grassroot level. When a new employee joins the company, the first thing s(he) learns is
how to follow Standard Operating Procedures (SOPs). If SOPs are in place, they will streamline
operations and save the business a significant amount of time. Because time is money in today's world,
this is an excellent tool for saving money.

https://doi.org/10.5281/zenodo.6420967
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 56

Theme 10: How should we speed up the process of establishing a supportive safety culture at
work?

Implementation entails building empathy for each other's safety on site, which necessitates sensitizing
the workforce to at-risk behaviors and constant spot-corrections. The most effective method is to
increase the number of qualified observers and make on-the-spot corrections during every shift. We
need to set up a way for employees to be able to say what they think. Without their help, the safety
culture movement won't move forward.

Regular management intervention, as well as sensitizing the workforce during daily toolbox talks and
meetings with a friendly approach, as well as observing safety behaviors on site, has accelerated the
safety culture of companies. Organizations should integrate HSE as a planned intervention into their
programs to make a safety culture a way of life. The trained behavioral safety observer is not limited to
the site; instead, he or she should make one observation daily, both on and off site, to make society
safer.

3. CONCLUSIONS AND RECOMMENDATIONS

To set the foundation for continuous improvement, a systematic process of assessment, identification of
strengths and weaknesses, deployment of focused interventions, and learning from the results is
required to change the safety culture (Ravi et al., 2021). The applications of psychology have
significantly improved site safety culture, as has the rigorous monitoring of behavioral risk patterns,
which is backed up by HSE systems and a strong commitment from management. Although there are
HSE standards in place, there is still a need to prevent fatalities on construction sites, for example.

Occupational safety and health practitioners must continue to learn in order to meet current regulations.
Multiple training approaches, particularly for junior safety practitioners, may stimulate competency
growth (Ishimaru et al. 2020). Because safety experts don't get enough basic training at college, there
have always been gaps in their skills, making it hard for them to learn more on the job, gain more
experience, and improve their performance.

When leadership commits to safety, every employee leads the firm to a culture of no hurt, no harm.
Even though underreporting is a dark reality and has always bothered people in the field of OHS,
strong monitoring (inspections) and a comprehensive database to figure out what to do and how to do it
are needed.

For industry and social science professionals, looking at reality through behavioral criteria is critical.
For long-term changes in safety culture, everyone involved must correct at-risk behaviors and apply
behavioral safety concepts and applications. These include psychological safety perspectives that go
beyond work content to include broad aspects of employees' personal circumstances and experiences
(Edmondson and Mortensen, 2021), as well as psychosocial risks that affect psychological health (ISO
45003, 2021).

Firms trying to change a safety culture focus too much on tools and processes, but they don't pay
attention to how employees think about their own safety and how that affects their behavior and
interactions (Hortense de la Boutetière, 2019). Business excellence is impossible to achieve without a
strong safety culture, and safety-time appears to be the best indicator of how much of a safety culture
has been developed in the organization by its employees through daily contributions to an observation
round. The active safety time of observations and spot-corrections can be correspondingly equated to

https://doi.org/10.5281/zenodo.6420967
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 57

the probability of injuries and mishaps on site (Kaila, 2020). The absence of a supportive safety culture
at work places could hurt the company; thus, it is important to get it quickly.

Individuals, companies, and the economy all suffer when job dangers are overlooked. As a result,
initiatives to promote a safety culture would save all of these in the long term, benefiting society as a
whole. When safety culture becomes a value system, and everyone values and speaks up for one
another's safety, it is deemed successful (Choueiri (2021).

In conclusion, we need to build and form safety culture implementation strategy groups at the national
or state level that can be taught to extend the training and implementation throughout industry and
society. This will benefit income tax, insurance, and contractors, as well as any other government or
commercial groups. All profits lost as a result of events, fires, fatalities, and accidents will be
conserved and distributed to the general public in some way.

REFERENCES
Barnes, Emily (2021, June, 15).NSC National Safety Month 2021: Week 3 – It’s Vital to Feel Safe on the Job.
Retrieved on 25 Nov 2021. https://www.ehs.com/2021/06/nsc-national-safety-month-2021-week-3-its-vital-
to-feel-safe-on-the-job/
Burdick, Guy (2019, May 13). Big Growth and Change Expected for the Safety Profession. Retrieved on 22 Nov
2021. https://ehsdailyadvisor.blr.com/2019/05/big-growth-and-change-expected-for-the-safety-profession/
Choueiri, Elias M. (2021, December). Individuals, Organizations, and the Economy: All Suffer When Workplace
Hazards Are Ignored! WSJ Vol. XXX No. 4, 67. https://doi.org/10.5281/zenodo.5865442
Edmondson, Amy C. and Mortensen, Mark (2021, April 19). What Psychological Safety Looks Like in a Hybrid
Workplace? Retrieved on 30 Nov 2021. https://hbr.org/2021/04/what-psychological-safety-looks-like-in-a-
hybrid-workplace
Hortense de la Boutetière, Jean-Benoît Grégoire Rousseau, and Eion Turnbull (2019, May, 21). Overcoming
limiting mind-sets to improve safety. https://www.mckinsey.com/business-functions/people-and-
organizational-performance/our-insights/overcoming-limiting-mind-sets-to-improve-safety
Ishimaru, Tomohiro & Punpeng, Twisuk & Maiyapakdee, Chavinthorn & Ketsakorn, Arroon & Fujino,
Yoshihisa & Hara, Kunio. (2020). Survey of the necessary competencies and proficiency of safety officers in
Thailand. Industrial Health. 58. 10.2486/indhealth.2019-0205.
ISHN (2021a, October 7). How does your safety culture impact workers? Retrieved on 22 Nov 2021.
https://www.ishn.com/articles/113124-how-does-your-safety-culture-impact-workers
ISHN (2021b, December 2). A young pro and his mentor address safety culture engagement - delivering safety
excellence. Retrieved on 29 Nov 2021. https://www.ishn.com/articles/113172-a-young-pro-and-his-mentor-
address-safety-culture-engagement
ISO 45003:2021 (2021). Occupational health and safety management - Psychological health and safety at work -
Guidelines for managing psychosocial risks. https://www.iso.org/obp/ui/#iso:std:iso:45003:ed-1:v1:en
Keenan, Paul Michael (2017, March 11). Spiritual Vulnerability, Spiritual Risk and Spiritual Safety - In Answer
to a Question: ‘Why Is Spirituality Important within Health and Social Care?’ at the ‘Second International
Spirituality in Healthcare Conference 2016: Nurturing the Spirit.’ Trinity College Dublin, The University of
Dublin. Religions 2017, 8, 38; https://www.mdpi.com/2077-1444/8/3/38
Kaila, H. L. (2020). Planned intervention of behaviour-based safety (BBS) helps organisations improve their
safety time arithmetic. World Safety Journal, 29(3), 34-41.

https://doi.org/10.5281/zenodo.6420967
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 58

Kaila, Harbans Lal (2021a). Behavioural Benchmarking as a Robust Tool of Total Safety Culture in Industries.
The Journal of Insurance Institute of India, July-Sept 2021, Vol. IX (1), 82-88.
Kaila, Harbans Lal (2021b). Why Industry is Losing Faith in HSE Awards?:Professionals Rethink. IJTD, 51(1),
82-90.
Kaila, Harbans Lal (2021c). Behavioural Economics of Safety Culture Management in Companies. OPUS, 12
(2) 2021, 21-42. http://www.publishingindia.com/opus/
Halligan, Michelle & Zecevic, Aleksandra & Kothari, Anita & Salmoni, Alan & Orchard, Treena. (2013).
Understanding Safety Culture in Long-Term Care: A Case Study. Journal of patient safety. 10.
10.1097/PTS.0b013e31829d4ae7.
MySafetySign (2021). Is management the biggest roadblock to safety? Retrieved on 29 Nov 2021.
https://www.mysafetysign.com/blog/roadblock-to-safety-programs/
Paoletta, David (2020, Jul, 1). Retrieved on 29 Nov 2021. Core Attributes of a World-Class Safety Culture.
https://safesitehq.com/safety-culture/
Pettinger, Chuck (2020, August, 25). 5 Ways to Build a Culture of Safety for the Long Term. Retrieved on 23
Nov 2021. https://www.ehstoday.com/safety-leadership/article/21140054/5-ways-to-build-a-culture-of-
safety-for-the-long-term
Ravi, D., Tawfik, D.S., Sexton, J.B. et al. (2021). Changing safety culture. J Perinatol. 41, 2552–2560.
https://doi.org/10.1038/s41372-020-00839-0
SHRM (2020, September 8). Accommodating Religion, Belief and Spirituality in the Workplace. Retrieved on
24 Nov 2021. https://www.shrm.org/resourcesandtools/tools-and- samples/toolkits/pages/
accommodating-religion,-belief-and-spirituality-in-the-workplace.aspx
Shyam Sundar, K.R. (2022, Feb, 14). A dipping graph in occupational safety and health. The Hindu.
https://epaper.thehindu.com/Home/ShareArticle?OrgId=GB59GCH21.1&imageview=0
Urbach, Tina and Fay, Doris (2020, February, 21). Leader Member Exchange in Leaders’ Support for Voice:
Good Relationships Matter in Situations of Power Threat. https://doi.org/10.1111/apps.12245
Velas, A., Halaj, M., Hofreiter, L. et al. (2021). Research of security and safety culture within an organization.
The case study within the Slovak Republic. Secur J (2021). https://doi.org/10.1057/s41284-021-00291-5
Vos, Jaime (2021, April 12). Empower your employees to create a culture of safety: here’s how. Retrieved on 27
Nov 2021. https://www.aem.org/news/empower-your-employees-to-create-a-culture-of-safety-heres-how
Williams, Oliver (2021, Sep., 1). Strengthen your safety culture. Retrieved on 28 Nov 2021.
https://www.ioshmagazine.com/2021/09/01/strengthen-your-safety-culture

Author
Harbans LAL earned a Master's degree in Psychology from Guru Nanak
Dev University and a Ph.D. from Tata Institute of Social Sciences, Mumbai.
He has been at SNDT Women’s University and the Central Labor Institute,
Mumbai, for over 28 years. He represented India at Conferences in New
York, Berlin, Muscat, Rome, New Zealand, Japan, London, Dubai, Cairo,
and Sydney. He is the Editor of the Journal of Psychosocial Research, and
serves as Director of the Forum of Behavioral Safety. He has conducted over
1000 behavioral safety programs for the industry.

https://doi.org/10.5281/zenodo.6420967
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 59

World Safety Journal

A peer-reviewed journal,
published by the World Safety Organization

Journal Homepage:
https://worldsafety.org/wso-world-safety-journal/

Public Transportation Infrastructure Projects and Financial Risks

Elias M. Choueiri
WSO Board Member and Liaison Officer to the United Nations

KEYWORDS ABSTRACT
Financial Risks When implementing a public transport infrastructure project, there are always a
Public Transport number of financing risks that need to be addressed before any such project is
Infrastructure Projects embarked upon. One of these risks concerns traffic demand/revenue risk – the
SkyWay envisaged public transport system needs to attract patronage; this requires private
Beirut
car users to make the switch to the alternative mode of transport. This would also
Lebanon
be the case for the various public transport infrastructure projects that are being
investigated to alleviate road traffic congestion in the Greater Beirut Area (GBA),
Lebanon, as alluded to in this article.

1. INTRODUCTION

S
uccessful private-sector funded public transport infrastructure projects are those projects that are
able to meet financial commitments and provide private participants their required rates-of-
return. The ability to meet financial commitments is a direct result of the establishment of proper
financial and project structures. Establishing the proper structures results from identifying key project
risks, then creating structures and policies that mitigate those risks.

2. PRINCIPAL RISKS IN FINANCING PUBLIC

TRANSPORT INFRASTRUCTURE PROJECTS

The principal risks involved in financing public transport infrastructure projects can be categorized into
(Estache & Strong, 2000; Schwartz et al., 2006):
• construction-phase dangers; and
• start-up and operating phase risks.

2.1 Construction-related dangers

During the construction phase, the major risks are:

• Delays in project completion and the commencement of project cash flows can result in an
increase in total costs through higher capitalized interest charges. It may also affect the
scheduled flow of project revenues necessary for debt servicing costs and operating and
maintenance expenditures.


Corresponding Author: [email protected]
https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 60

• cost overruns with an increase in the capital needed to complete construction. For a variety of
reasons, including incorrect engineering and design, increased material and labor costs, and
project start-up delays;
• Contractor insolvency or lack of experience: The main contractors and key subcontractors
should have the experience and reputation, as well as the financial, technical, and human
resources, to complete the project on time and within budget.

Further, the risk of non-availability of materials or equipment for construction or operation exists,
especially in many developing countries. This is especially true with respect to specialized equipment.
Also, environmental and land risks may cause delays. Public transport infrastructure projects can have
a substantial environmental impact. Such projects frequently attract strong opposition from community
and environmental groups over issues of pollution, congestion, and visual impact. Similarly, land
acquisition can be a protracted process with the potential for extensive legal delays, particularly in
developing countries. Furthermore, the cost of land acquisition can become a major factor where land
values have risen rapidly or are subject to speculative activity over which the project developer has no
control. In these cases, agreement on some form of cost ceiling may be necessary in the concession
contract.

2.2 Start-up and operating phase risks

The major risks for public transport infrastructure projects during the start-up and operating phase are:
• Technology risks: project financing participants cannot ignore new technologies since they can
significantly improve the profitability of a project. The use of obsolete technology may
adversely affect a project.
• Financial risks (interest rate and foreign exchange rate fluctuations): since the advent of
financial crises in emerging markets, few projects are able to generate returns-on-investment
sufficient to attract private capital. This suggests that until macro-economic risk premiums
decline and traffic demand is more established, only a limited set of projects will be undertaken
without substantial government support. The financial crisis may force many programs to slow
down and force debt restructuring of many of the existing concessions. Since the risk of bailouts
is high, there is a need for more secure financing structures.

Another financial risk is driven by the impact of fluctuations in foreign exchange rates on the value of
the business. This is a major issue for some projects, where revenues are commonly in local currency
and adjustments for inflation and foreign exchange rates may lag or encounter political opposition.
• Force majeure risks: force majeure refers to risks beyond the control of either the public or the
private sector, such as flooding or earthquakes, which impair the ability of the project to earn
revenues. The rule is that if there is a risk of force majeure, the contract should say what
happens if that happens.
• Regulatory and legal risks: regulatory risks stem from the weak implementation of regulatory
commitments built into concession contracts but also in laws or other legal instruments relevant
to the value of the transaction. The question to be asked is whether the regulator will exercise its
authority and responsibilities over prices, public obligations, and competition rules, as well as
other rules that are specified in the contracts and influence the value of the business. Because,
even if regulatory rules are clear enough, they are only as effective as the regulators can be.

Legal risks entail the fact that, typically, project financing structures cover periods of ten years or more.
However, over that time period, the relevant legal and regulatory environment is likely to change
substantially. The rules governing the financial consequences of these changes between the

https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 61

government, users, and operators are critical, but they are frequently overlooked. The rules must allow
for changes to the contract terms during the time that the project is being funded.
• Governments generally agree to compensate investors for political risks, although in practice,
justifications for government actions may be cited to delay or prevent such payments. Thus, private
investors generally assume the risks associated with dispute resolution and the ability to obtain
compensation should the government violate the concession agreement. The issue of meeting financial
obligations while disputes are resolved may be achieved through a requirement for debt servicing
reserves, contingency funds, or standby financing.

Political risks concern government actions that affect the ability to generate earnings, such as actions
that terminate the concession, the imposition of taxes or regulations that significantly reduce the value
of the business for investors; placing restrictions on the ability to collect or increase tariffs; precluding
contract disputes from being resolved in reasonable ways;

2.3 Traffic demand/revenue risks

Unlike project financing in other sectors, take-or-pay or fixed-price contracts are typically not available
in public transport infrastructure projects, so traffic demand risk is a major issue in virtually all of these
projects. Even when there is a reasonable level of confidence in forecasts, traffic demand can be
dramatically affected by competition from other modes or facilities, changing usage patterns, and
macro-economic conditions. These inter-related issues, over which the project sponsor often has little
or no control, are very difficult to predict and represent a major risk to financing. In particular, traffic
demand forecasting during the early years can be quite subjective. To the extent that these risks are
driven by economic conditions, there is a potential role for the government to play in risk sharing,
either through traffic demand, revenue guarantees, or other forms of support.

3. ATTRACTING PRIVATE-CAR USERS TO

BECOME A PUBLIC TRANSPORT PATRON

For any newly implemented public transport system, it is of great importance to attract patronage. This
would also be the case for those that are under study to solve the dire road traffic congestion situation
in the Greater Beirut Area (GBA) in Lebanon, where currently no efficient and reliable public transport
system is in place. Most commuters use their private cars to get to/from their places of work or
education. In this respect, it is worth noting that over 24% of employment opportunities in Lebanon are
located along the Beirut-Jounieh (Tabarja) corridor, and that the majority (78%) of commuters
travelling on this route use their private cars; the remaining 22% use taxi (van) services, and (mainly
private sector) minibus/bus services that do not really provide scheduled services. All this road traffic
on the Northern Corridor, in addition to that on the Eastern and Southern Corridors, lies at the source of
the dire road traffic congestion situation in Beirut, which is such that up to 70% of commuter travel
time is lost in delays due to the sa(i)d road traffic congestion situation. The average reported inter-
modal road traffic speed is 11 km/h (calculated as the weighted average speed across all modes), with
an average speed of 9.4-13.5 km/h for private cars, 6.5-9.3 km/h for buses, 7.5-10.8 km/h for
minibuses, and 8.6-12.3 km/h for taxis. As a result, every day, commuters are forced to spend hours
stuck in traffic jams and in their cars, struggling to somehow reach their respective destinations.

As noted in previous articles (e.g. Choueiri, 2014, 2018, 2020, 2021a), studies to implement an
efficient and modern public transport system to alleviate road traffic congestion in the Greater Beirut
Area (GBA) have been underway for several years now, but are struggling to make much progress due
to the various political, economic, financial, and other crises Lebanon is currently plagued with.

https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 62

Among these studies, railways are also being looked at, such as revitalizing (using the already existing
rights-of-way) railway services on the routes (see Map):
• Beirut-Jounieh-Tabarja-Maameltein (Tripoli); and
• Tripoli-Abboudiyeh.

3.1 The SkyWay option

Another option being considered is to build an elevated SkyWay system between Beirut Rafic Hariri
International Airport and Casino du Liban (Jounieh), with intermediate stations at strategic locations
(Aoun et al., 2019; Choueiri, 2021b), which could achieve significantly shorter travel times than a
private car (see Figures 1 and 2). Travelling by private car from Beirut-Rafic Hariri International
Airport to Casino du Liban in Jounieh, a distance of 49.7 km, takes about 1 hour and 50 minutes.
However, as can be observed from the table below, when travelling by the SkyWay system (average
travel speed of 100 km/h), which would have a shorter route of 40 km (10 km shorter than the road
alternative), the travel time would be just 30 minutes (this figure takes into account a travel speed of
approximately 5 km/h when passing through the intermediate stations, as people would embark and
disembark, which would take about 1 minute per station), see Table 1.

Figure 1. Beirut-Jounieh SkyWay Investigated Alignment

and Station Locations on Google Earth Basemap

https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 63

Figure 2. GIS Simulation of the SkyWay Investigated Alignment and Station Locations

In order to meet Ynitskiy's (or Unitsky’s) standards, a minimum radius of 1,000 meters was considered
for the entire project, resulting in an acceptable speed of 100 km/h along the entire alignment on
straight and curved sections (Aoun et al., 2019). In the Unitsky String Technologies (UST) system
architecture, a distinction was drawn between two types of support. Every one kilometer, an anchor or
tensioning support is employed; the second type is an intermediate support, which was employed to
alleviate the sag effect/vertical deflections on the structure. Based on the clearance required from the
bottom of the unibus and the travel path below, the typical height of each support erected on the ground
is 10 meters above the roadway level and 18 to 25 meters above sea level. There are 43 anchor supports
and 1,285 intermediate supports in total on the planned route.

https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 64

Table 1. Travel Time Comparisons between Personal Vehicle trips and the SkyWay Technology
Origin-Destination Length - Time - Personal Vehicle Length - Time -
Personal (mins) UST UST
Vehicle (km) (km) (mins)
Beirut Airport - Raouche 10.6 27 10.037 6
Raouche - Zeitouna Bay 5.5 23 4.441 3
Zeitouna Bay – City Mall 10.2 23 7.363 4
CityMall - Le Royal 9.2 11 7.310 4
Le Royal - Military Club 6.2 10 5.736 3
Military Club - Casino 8.0 17 5.183 3
Total 49.700 111 (1 hr 51 mins) 40.070 30 mins

3.1.1 Cost Estimation

Table 2 shows a summary of the project's costs.

Table 2. UST Construction Cost

Item Unit Price Number of Units Total Cost
(USD) (Million USD)
610UB152 Beam 1,555.50 1328 pieces 2.07
502CHS12.1 Column 1,656 1328 pieces 2.20
1000x7000 Pile 1,705 1328 pieces 2.26
35 mm Cable 37 480840 m 17.79
Concrete 75.50 2172.89 m³ 0.16
Materials (Extra) 750 1328 pieces 1.00
Construction Labor 2,996,540 40.07 km 120.07
Terminal Stations 4,167,251 2 pieces 8.33
Intermediate Stations 100,014 5 pieces 0.50
Traffic/Site Management 1,048,789 40.07 km 42.02
Design and Commissioning 111,279,093.36 9% 10.02
Land Acquisition - - 5.28
Total - - 211.70

3.1.2 UST Feasibility Study Results

When public transportation is well-established, driving a private automobile during rush hour becomes
a luxury rather than a necessity. As a result, people who choose this level of luxury will pay a higher
price in tolls, parking fees, and congestion costs. The following are the key findings from the feasibility
study for the UST Project:

3.1.2.1 Ridership Potential

This system may create an average yearly revenue of $3 million if ridership is between 3,000 and 4,000
per day, which is comparable to an average of 1,280,000 people per year, depending on fare price.

https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 65

3.1.2.2 Implementation Costs

A UST project system costs $211.7 million in total (see Table 2), or $5.3 million per kilometer. As a
result, when compared to the expenses of other forms of transportation, this system is extremely cost-
effective.

3.1.2.3 Energy Consumption

The UST cableway has a low energy demand of 1.8 million kilowatts per year. Users who shift away
from automobile trips result in lower GHG (Greenhouse Gas) and carbon emissions. The more the
system expands, the higher the commuter trips that result in lower additional emissions.

3.1.2.4 Comparative Analysis with Other Modes of Transport

High speed, high efficiency, and low rollout cost are the fundamental characteristics of a UST system.
Figure 3 highlights comparisons with various modes of transportation in terms of capital costs,
pollution, operational costs, and traffic accident rates.

Figure 3. Comparison between Transport Modes Regarding:

a) Capital Expenses, b) Environmental Pollution,
c) Operating Expenses, and d) Traffic Accident Rate

https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 66

SkyWay vs. Cableway (Dunayeva & Vinakurava, 2019)

The principle of the cableway is the movement of the cable, on which the cars are fixed, between two
points. In SkyWay, the vehicle (electric rail vehicle) is not moving along a cable that requires regular
replacement every 6–8 years, but on a steel profile. When moving at the same speed, energy
consumption in the string railroad will be 5–6 times lower compared to the cableway.

SkyWay surpasses the cableway in a range of parameters – operating costs, service life, safety, speed,
performance. It is worth noting that SkyWay, in contrast to the cableway, has various types of tracks,
making it possible to combine them, and a wide range of vehicles.

Table 3. SkyWay vs. Cableway

Characteristics Cableway SkyWay
Speed 21.6 km/h (6 m/s) Up to 150 km/h (in city)
Track length Up to 10 km Unlimited
Principle of Outboard motor moves both cable and Autonomous steel-wheeled
movement cars
Turning capability No Yes
Service life Complete replacement of cables each 6–8 Track structure: 50–100 years; Rolling stock:
years 25 years
Safety Low High
Comfort No climate control Climate control
Cost effectiveness Low High

SkyWay vs. LRT (Dunayeva & Vinakurava, 2019)

Compared to LRT, SkyWay vehicles’ capacity is much more attractive than that of the Light Rail
Transit (LRT).

Table 4: SkyWay vs. LRT

Characteristics LRT SkyWay

Coupled unibuses: up to 240 people on the upper track
Train: from 4–5 cars from
Rolling and up to 175 people on the lower track. Platform
up to 100 passengers
stock length: up to 50 m
each.
capacity
Platform length: minimum
150 m
Headway From 1–2 minutes From 20–25 seconds
Safety Medium High

SkyWay vs. Monorail (Dunayeva & Vinakurava, 2019)

Compared to monorail, SkyWay wins primarily in terms of speed. In addition, monorail requires a very
massive overpass, with high material capacity.

https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 67

Table 5: SkyWay vs. Monorail

Characteristics Monorail SkyWay

Track length Up to 50 km Unlimited
Train: from 4 cars for up to
Coupled unibuses: up to 240 people on the upper track
Rolling stock 80 passengers and more
and up to 175 people on the lower track. Platform length:
capacity each.
up to 50 m
Platform length: minimum
100 m
Maximum 70 km/h 150 km/h
speed
Elevated pneumatic-tired Autonomous steel-wheeled vehicles
Principle of
transport
movement
Headway 5–20 minutes From 20–25 seconds
Track structure: 50 years;
Track structure: 50–100 years; SkyWay rolling stock: 25
Service life Rolling stock: 10–15 years
years

3.1.2.5 Funding Opportunities

Because the UST system is both creative and cost-effective, it is an excellent contender for funding.
Local improvement districts, public and private partnerships, as well as state, federal, and local transit
enhancement programs, could all be feasible funding sources.

3.1.2.6 Traffic Assessment Impacts

By 2050, the urban population in Lebanon is expected to have grown to 87 percent of the overall
population (Khoury, 2017). Such rapid expansion would result in increased traffic congestion,
increased traffic accident risks, and unreliable public transportation systems. Furthermore, the
domination of private passenger automobiles would result in overcrowding on current roadways, higher
GHG emissions and pollution levels, and higher total mobility costs. Without UST, all existing public
transportation modes in Lebanon would result in just unorganized traffic, which would have a negative
impact on global warming, mobility, cost safety, and social productivity.

By all accounts, the UST System has the potential to attract a large number of travellers, as 29% of
people now use public transportation (Khoury, 2017). Traffic volumes would be reduced by 3,680
vehicles per hour during AM times (4417 passengers) and 2,925 vehicles per hour (3,510 passengers)
during PM periods, based on a carrying capacity of 12,200 per day and an occupancy of 1.2 individuals
per car. The number of cars would be reduced by 8,362 (10,034 passengers) during AM periods and
6,644 (7,973 passengers) during PM periods if 65.88 percent of residents considered using the UST
system.

3.1.3 Project Impact Assessment

Adopting some form of public transportation is, without a doubt, the most consistent, safe, and
convenient choice. However, because Lebanon's road space and land area are limited, allocating a bus
lane would almost surely worsen traffic congestion. In this regard, the UST elevated railway
transportation system avoids issues like road space and uses only 1.8 kilowatts per year, demonstrating
its excellent environmental friendliness (Yunitskiy, 2015). The O-D (Origin-Destination) speeds clearly
https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 68

illustrate that this technology is far faster than a car and easily avoids all traffic bottlenecks. In terms of
cost, the UST is a viable option when compared to other modes of transportation, costing roughly $8
million per 1 km of construction with a 50% contingency. Furthermore, UST's zero-tolerance policy is
highly effective.

3.1.4 Project Disadvantages and Limitations

Every public transportation system offers a set of advantages. However, some disadvantages may arise
as a result of the UST system's implementation; these are listed below:
• The most challenging issue in any public transportation system is funding and regulatory
policies.
• Pile foundations necessitate a high level of precision, which could drive up building costs.
• Relevant stations necessitate site space and regulatory approvals for land acquisition.

As a result, more comprehensive soil research and geotechnical studies would be required to reduce the
aforesaid disadvantages, and regulatory requirements would need to be examined to avoid any on-site
operational risks.

3.2 SkyWay – a commuter survey

With respect to the SkyWay option, a survey was held among 306 current commuters in Beirut aimed
at gathering their viewpoint on their current use or non-use of public transport and to acquire an insight
into whether they would be inclined to use the SkyWay system should it become available. Amongst
other things, the results of this survey yielded the following (Aoun et al., 2019):
• 238 (78%) of the respondents regularly used the private car as their primary mode of transport,
148 (62.18%) of them would consider shifting from travelling by private car to travelling by
SkyWay system should it become available;
• 61 (20%) of the respondents used some form of public transport (taxi (van) services, and
(mainly private sector) minibus/bus services), and a negligible proportion used other modes of
transport (bike and motorcycle);
• 203 of the respondents earned less than $1,500 per month, 116 of them spent $100-$300 (7%-
20% of their monthly income) on transport each month. From these 116 respondents, 96
(82.76%) used the private car as their prime mode of transport. Further, 56 of the respondents
travelled by bus. Of those travelling by bus, 52 (92.8%) had an income of less than $1,500 per
month (39 (75%) of them spent less than $100 on transport each month). These figures seem to
suggest that, despite their low incomes, the respondents prefer to use their private cars;
• 41.07% of the 56 respondents who travelled by bus spent about 30-60 minutes from home to
their respective places of work/education. Of the 238 respondents who travelled by car, 108
(45.38%) also spent 30-60 minutes to travel from home to their places of work/education. Both
modes of transport get stuck in the same traffic jams;
• 111 (33%) of the respondents ranked travel time to reach one’s destination as their first priority
for choosing a mode of transport. For this reason, 81 (73%) of them would consider using the
SkyWay system which, as noted earlier, would offer very favorable travel times, as an
alternative option. Thus, travel time would be a major criterion for using public transport.

It has been estimated that the SkyWay system which, as noted earlier, could offer more favorable travel
times than the private car on the route Beirut-Jounieh (as it passes over all traffic congestion obstacles),
could attract some 3,000-4,000 passengers per day, which would be equivalent to an average of
1,280,000 passengers per year (Aoun et al., 2019). In view of the low car occupancy rate of 1.2

https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 69

persons/car in the Greater Beirut Area (GBA) (Kadi, 2016), this could lead to a significant reduction in
the number of cars circulating on the roads.

3.3 Attracting private car users to become public transport patrons

For any public transport system to be conceptualized to accommodate a reasonable level of ridership on
a given corridor, it is essential that the system specifically understands, targets, and adapts itself to the
current private car users. Once an objective has been established, it can then be carried out through
respective levels of subsidy as well as fare and price discrimination mechanisms.

To attract current private car users, the majority of whom would be middle-class professionals, to
patronize the public transport system, it should be recognized that one of the key variables, besides the
price of tickets, would be the location of stations. Also, in suburban areas, the stations would need to
have adequate parking facilities as the majority of commuters would need to park their cars at the
station and then use the public transport system to commute into the city. As for stations in Beirut, it
would be very important to address the issue of transport mode transfer, as well as to consider the
walking distance from the station to the Central Business District or other destinations in the city. For
instance, requiring a commuter to use three modes of transport to get to his/her place of work would
likely diminish ridership. Noteworthy in this respect are the two following comments that were made
by respondents to the SkyWay survey (Aoun et al., 2019):
• If SkyWay coincides with my time and destination, it will be my choice; otherwise, I will use
my car.
• There is also the comfort criterion! If it is a hassle to use SkyWay, then I would still consider
using my car.

4. FINAL REMARKS

In the various articles that have been published in this journal over the years, it has become quite clear
that Lebanon, and especially the Greater Beirut Area (GBA), is in dire need of a modern, reliable, and
efficient public transport system that can take commuters to their places of work and education, as well
as others to their respective destinations, in a more comfortable and environmentally-friendly manner
than when travelling by private car, which is so often characterized by being stuck in polluting traffic
jams. A well-established public transport system would greatly enhance the mobility and quality of life
of the people, as well as bring other societal and economic benefits. Once in place, people would
become more and more familiar and appreciative of its many benefits, thus turning many of them into
patrons of such a system.

REFERENCES

Aoun J., Choueiri E.M. & Otayek E. (2019). Alleviating traffic congestion in Lebanon with special emphasis on
the roadway section: Beirut-Jounieh. Advances in Transportation Studies, pp. 87–104.
Choueiri E.M. (2014). Alleviating traffic congestion by implementing LRT: studies underway for Beirut. REI,
Number 4, p. 9.
Choueiri E.M. (2018). Revitalising public transport in Lebanon – a matter of urgency: steps that are being
undertaken towards making it a reality (Part 1 of 2). REI, Number 3, pp. 14-16.
Choueiri E.M. (2020). An efficient public transport system could combat the serious road traffic congestion
situation in Beirut, Lebanon – a country that is currently facing a dire economic crisis. REI, Number 3, pp. 4-
7.
Choueiri E.M. (2021a). Road traffic air pollution in the Greater Beirut Area (GBA), Lebanon – impact on public
health: urgent need for public transport development, especially rail. REI, Number 1, pp. 13-16.

https://doi.org/10.5281/zenodo.6420979
World Safety Journal (WSJ) Vol. XXXI, No 1 Page 70

Choueiri E.M. (2021b). SkyWay: a possible solution proposed to alleviate the very dire road traffic congestion
and air pollution situation in the Greater Beirut Area (GBA), Lebanon. REI, Number 3, pp. 12-15.
Dunayeva, M. & Vinakurava, M. (2019). Skyway innovation technology for smart cities. Urban Transport XXV.
Estache A. & Strong J. (2000): The Rise, the Fall, and... the Emerging Recovery of Project Finance in Transport.
World Bank Institute.
Kadi S. (2016, December 4). Traffic congestion adds to Lebanon’s many woes. The Arab Weekly.
Khoury, R. (2017). Environmental and Social Impact Assessment (ESIA) For The Bus Rapid Transit (BRT)
System Between Tabarja and Beirut and Feeders Buses Services. Council for Development and
Reconstruction (CDR), Lebanon.
Schwartz G., Corbacho A. & Koranchelian T. (2006, June 30). Financing Transportation Infrastructure—
Potential Fiscal Risks of Innovative Financing Mechanisms. International Monetary Fund.
Yunitskiy.com, Transport Complex SkyWay in Questions and Answers : 100 Questions - 100 Answers. [online]
Available at: http://yunitskiy.com/author/2016/2016_98.pdf, 2015.

Author
Prof. Dr. Elias M. CHOUEIRI has been very active in academic and research
settings for over 35 years. He is the author/co-author of over 20 books and
booklets, and hundreds of refereed publications, technical reports, conference
presentations and newspaper articles. He has won more than 20 awards for his
scholarship, and has held faculty and managerial positions at several public and
private institutions in Lebanon and the USA. He is a member of the WSO Board
of Directors, and serves as WSO Liaison Officer to the United Nations. Besides,
he assumes the roles of Director of the WSO National Office for Lebanon,
Chairperson of the WSO Highway Transportation Committee, and Chairperson
of the WSO Transportation of Dangerous Goods Committee.

POST SCRIPTUM

Please note that the survey addressed in this article was conducted in 2017/2018, i.e. prior to the financial and
economic collapse of Lebanon that started in October 2019, and is still continuing this very day, exacerbated
even more by the current afflictions taking place in Ukraine. “Up to 90% of Lebanon’s wheat and cooking oil
imports come from Ukraine and Russia, as well as a large proportion of grain imports” (citation from news item
‘Lebanese fearful as fuel and wheat shortage deepens: The war in Ukraine has left cash-strapped Lebanon
scrambling for alternative sources of fuel and wheat’, Al Jazeera Live, 8 March 2022). Also, the Lebanese Pound
has lost some 90% of its value against the US Dollar, which also puts the income figures noted in the article in a
different light. Furthermore, as the price of gasoline has gone completely out of control, also exacerbated by
what is presently going on in Ukraine and the various restrictions that are in place, only a very small proportion
of people can afford to travel to their places of work/education (these days, the price for 20 liters of gasoline
stands at $300-00, i.e. $15-00 per liter, to put it into perspective).

Remaining hopeful that, one day, things will get back to normal, solving the issue of the dire road traffic
congestion situation in Beirut, either by implementing a ground-based railway system or an over-ground
SkyWay system, should remain high on the agenda. It is assumed that, once in place, commuters will soon see
the advantage of a modern, reliable and efficient public transport system that can take them to their places of
work/education in a passenger-friendly, relaxed and comfortable manner – quite the opposite from the stressful
daily battle to get to one’s destination.

https://doi.org/10.5281/zenodo.6420979
 World Safety Organization (WSO)
The WSO was founded in 1975 in Manila, The Republic of the Philippines, as a result of a gathering of over
1,000 representatives of safety professionals from all continents at the First World Safety and Accident
Prevention Congress. The WSO World Management Center was established in the United States of America in
1985 to be responsible for all WSO activities, the liaison with the United Nations, the co-operation with numerous
Safety Councils, professional safety/environmental (and allied areas) organizations, WSO International
Chapters/Offices, Member Corporations, companies, groups, societies, etc. The WSO is a non-profit, non-
sectarian, non-political organization dedicated to: “Making Safety a Way of Life … Worldwide.”

World Safety Organization Activities corporations, companies, and other entities paying the
WSO publishes WSO Newsletters, World Safety Journal, WSO Membership and/or WSO Certification fees for
and WSO Conference Proceedings. their employees.
WSO provides a network program linking various areas Members have access to WSO Newsletters and other
of professional expertise needed in today’s international member- ship publications of the WSO on the WSO
community. website, and may request hard copies by contacting the
WSO World Management Center. Subscription fees
WSO develops and accredits educational programs apply to certain publications.
essential to national and international safety and
establishes centers to support these programs. Members are entitled to reduced fees at seminars,
conferences, and classes given by the WSO. This
WSO receives proposals from professional safety groups/ includes local, regional, and international programs.
societies for review and, if applicable, submits them to When Continuing Education Units (CEUs) are
the United Nations for adoption. applicable, an appropriate certificate is issued.
WSO presents annual awards: The James K. Williams Members who attend conferences, seminars, and classes
Award, Glenn E. Hudson International Award, J. Peter receive a Certificate of Attendance from the WSO. For
Cunliffe Transportation Award, Concerned Citizen, individuals attending courses sponsored by the WSO, a
Concerned Company/Corporation, Concerned Certificate of Completion is issued upon completion of
Organization, Educational Award, WSO each course.
Chapter/National Office of the Year, and Award for
Achievement in Scientific Research and Development. Members receive special hotel rates when attending
safety pro- grams, conferences, etc., sponsored by the
WSO provides recognition for safety publications, films, WSO.
videos, and other training and media materials that meet
the WSO required educational standards. Membership
WSO establishes and supports divisions and committees The World Safety Organization has members who are
to assist members in maintaining and updating their full time professionals, executives, directors, etc.,
professional qualifications and expertise. working in the safety and accident prevention fields,
WSO has Chapters and National/International Offices including university professors, private consultants,
located throughout the world, providing contact with expert witnesses, researchers, safety managers, directors
local communities, educational institutions, and of training, etc. They are employees of multinational
industrial entities. corporations, local industries, private enterprises,
governments, and educational institutions. Membership
WSO organizes and provides professional support for
in the World Safety Organization is open to all
inter- national and national groups of experts on all
individuals and entities involved in the safety and
continents who are available to provide expertise and
accident prevention field, regardless of race, color,
immediate help in times of emergencies.
creed, ideology, religion, social status, sex, or political
Benefits of Membership beliefs.
WSO publishes the “WSO Consultants Directory” as a Membership Categories
service to its Members and to the Professional
Associate Membership: Individuals connected with
Community. Only Certified Members may be listed.
safety and accident prevention in their work or
WSO collects data on the professional skills, expertise, individuals interested in the safety field, including
and experience of its Members in the WSO Expertise students, interested citizens, etc. Affiliate Membership:
Bank for a reference when a request is received for Safety, hazard, risk, loss, and accident prevention
professional expertise, skill, or experience. practitioners working as full time practitioners in the
WSO provides a network system to its Members safety field. Only Affiliate Members are eligible for the
whereby professional assistance may be requested by an WSO Certification and Registration Programs.
individual, organization, state, or country or a personal Institutional Membership: Organizations, corporations,
basis. Members needing assistance may write to the agencies, and other entities directly or indirectly involved
WSO with a specific request, and the WSO, through its in safety activities and other related fields.
Membership and other professional resources, will try to Sustaining/Corporate Member: Individuals,
link the requester with a person, organization, or other companies, corporations, organizations or other entities
resource which may be of assistance. and selected groups, interested in the international effort
WSO provides all Members with a Membership to “Make Safety A Way of Life ... Worldwide.”
Certificate for display on their office wall and with a The WSO Membership Application is included on the
WSO Membership Identification Card. The WSO awards following pages and is also available on the WSO
a Certificate of Honorary Membership to the website: https://worldsafety.org/quick- downloads/
WSO – Application for Membership
 WSO – National Offices
WSO National Office for Algeria WSO National Office for Lebanon
c/o Institut des Sciences et de la Technologie (I.S.T.) c/o Ministry of Transport
attn.: Mr. Ferhat Mohia, Director attn.: Dr. Elias M. Choueiri, Director
contact: [email protected] contact: [email protected]

WSO National Office for Australia WSO National Office for Myanmar
c/o Curtin University of Technology c/o Win Oshe Services Co., Ltd
attn.: Dr. Janis Jansz, Director attn.: Mr. Win Bo, Director
contact: [email protected] contact: [email protected]

WSO National Office for Austria WSO National Office for Nigeria
c/o Payesh System Mehr Engineering Company c/o DanaRich Creative Concept, LTD
attn.: Dr. Majid Alizadeh, Director attn.: Mr. Soji Olalokun, WSO-RSD, Director
contact: [email protected] contact: [email protected]
website: worldsafety.org.ng
WSO National Office for Cameroon
c/o Cameroon Safety Services WSO National Office for Pakistan
attn: Mr. Clement B. Nyong, Director c/o Greenwich Training & Consulting
contact: [email protected] attn.: Mr. Tayyeb Shah, Director
contact: [email protected]
WSO National Office for Canada
c/o Apex One Management Group WSO International Office for Philippines
attn.: Mr. Michael Brown, Director attn.: Engr Alfredo A. De La Rosa Jr., Director
contact: [email protected] | contact: [email protected]
[email protected]
website: worldsafetycanada.ca WSO National Office for Saudi Arabia (KSA)
c/o The Academy of Sciences for Medical Education
WSO National Office for Ghana attn.: Mr. Rocky Binuya, Director
c/o Ghana National Fire Service contact: [email protected] |
attn.: Mr. Peter Oko Ahunarh, Director [email protected]
contact: [email protected] website: https://aos-ksa.com/en

WSO National Office for India WSO National Office for United Arab Emirates
c/o Indian Society of Safety Engineers (I.S.S.E) (UAE)
attn.: Mr. T. Shankar, Director c/o Tatweer Industrial Inspection & Training
contact: [email protected] Services LLC
attn.: Miss Nazya Robin, Quality Manager &
WSO National Office for Indonesia Director
c/o Prosafe Institute contact: [email protected]
attn.: Mr. Soehatman Ramli, Director
contact: [email protected] WSO National Office for Vietnam
c/o Safety Training & Consulting Limited
WSO National Office for Iran attn.: Mr. Binh Pham, WSO-CSI(ML), Director
c/o Payesh System Mehr Engineering Company contact: [email protected]
attn.: Mrs. Fatemeh Gilani, Director [email protected]
contact: [email protected]
website: worldsafety.org.vn
WSO National Office for Iraq
c/o NAYA Engineering Services & Training
attn.: Dr. Eng. Khaldon Waled Suliman, Director
contact: [email protected]
Published by the WSO National Office for Lebanon
www.worldsafety.org
[email protected] | [email protected]

© 2022