Open Eng.

2015; 5:75–88

Vision Article Open Access

Ahmed K. Noor*

Potential of Cognitive Computing and Cognitive

Systems
Abstract: Cognitive computing and cognitive technologies
are game changers for future engineering systems, as well
1 Introduction
as for engineering practice and training. They are ma-
The history of computing can be divided into three eras
jor drivers for knowledge automation work, and the cre-
([1, 2], and Figure 1). The first was the tabulating era, with
ation of cognitive products with higher levels of intelli-
the early 1900 calculators and tabulating machines made
gence than current smart products.
of mechanical systems, and later made of vacuum tubes. In
This paper gives a brief review of cognitive computing
the first era the numbers were fed in on punch cards, and
and some of the cognitive engineering systems activities.
there was no extraction of the data itself. The second era
The potential of cognitive technologies is outlined, along
was the programmable era of computing, which started
with a brief description of future cognitive environments,
in the 1940s and ranged from vacuum tubes to micropro-
incorporating cognitive assistants - specialized proactive
cessors. It consisted of taking processes and putting them
intelligent software agents designed to follow and inter-
into the machine. Computing was completely controlled
act with humans and other cognitive assistants across the
by the programming provided to the system. The third era
environments. The cognitive assistants engage, individu-
is the cognitive computing era, where computing technol-
ally or collectively, with humans through a combination
ogy represented an intersection between neuroscience, su-
of adaptive multimodal interfaces, and advanced visual-
percomputing and nanotechnology.
ization and navigation techniques.
In a little more than a century computing shifted from
The realization of future cognitive environments requires
numbers to data then from data to knowledge. The shift
the development of a cognitive innovation ecosystem for
was not about having one system replacing the other but
the engineering workforce. The continuously expanding
enriching it. Programmable systems enabled the creation
major components of the ecosystem include integrated
of data by processing numbers, and cognitive computing
knowledge discovery and exploitation facilities (incor-
allowed making sense of data. Sense is what stands be-
porating predictive and prescriptive big data analytics);
tween raw data and actionable data.
novel cognitive modeling and visual simulation facilities;
Cognitive computing has attracted attention since 2011
cognitive multimodal interfaces; and cognitive mobile and
when the IBM Watson computer (of the IBM DeepQA
wearable devices. The ecosystem will provide timely, en-
project) played against two champions of the US game
gaging, personalized / collaborative, learning and effec-
show Jeopardy and won. Watson was able to respond di-
tive decision making. It will stimulate creativity and inno-
rectly and precisely to natural language prompts with rele-
vation, and prepare the participants to work in future cog-
vant responses. It had access to 200 million pages of struc-
nitive enterprises and develop new cognitive products of
tured and unstructured information consuming four ter-
increasing complexity.
abytes of disk storage.
Keywords: Cognitive computing, Cognitive systems, Cog- Whereas in the programmable era, computers essentially
nitive products process a series of ’if then what’ equations, cognitive sys-
tems learn, adapt, and ultimately hypothesize and suggest
DOI 10.1515/eng-2015-0008
answers. With the advent of big data, which grows larger,
Received August 27, 2014; accepted October 29, 2014 faster and more diverse by the day, cognitive computing
systems are now used to gain knowledge from data as ex-
perience and then generalize what they have learned in
new situations ([3] and Figure 2). They unlock the insights
that the new wealth of data generated holds. Delivering
*Corresponding Author: Ahmed K. Noor: Center for Advanced En- these capabilities will require a fundamental shift in the
gineering Environments, Old Dominion University, Norfolk, United
way computing progress has been achieved for decades.
States, E-mail: [email protected]
© 2015 Ahmed K. Noor, licensee De Gruyter Open.
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.
- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
76 | Ahmed K. Noor

Figure 1: Three Eras of Computing (based on [1]).

erated at a higher rate of speed than the human brain did.

2 Definition and characteristics of Its major asset is being able to accelerate the rate of learn-
cognitive computing ing in order to support humans in their work. Cognitive
computing and cognitive technologies can be considered
2.1 Definition of Cognitive Computing as the third phases of the AI evolution, from traditional
Artificial Intelligence (AI) to Artificial General Intelligence
Cognitive computing refers to the development of com- (AGI) to cognitive systems [5].
puter systems modeled after the human brain, which has
natural language processing capability, learn from experi-
ence, interact with humans in a natural way, and help in 2.2 Relation to Neural networks
making decisions based on what it learns [2–4]. All cogni-
tive computing systems are learning systems. They incor- Cognitive computing integrates the idea of a neural net-
porate embedded data analytics, automated management work, a series of events and experiences which the com-
and data-centric architectures in which the storage, mem- puter organizes to make decisions. Neural networks mimic
ory, switching and processing are moving ever closer to the the behavior of the human brain. Like the brain, multi-
data. Their way of processing massive amounts of data is layered computer networks can gather information and re-
neither linear nor deterministic. act to it. They can build up an understanding of what ob-
Originally referred to as artificial intelligence, researchers jects look or sound like. They contribute to the computer’s
began to use the modern term cognitive computing instead body of knowledge about a situation and allow it to make
in the 1990s, to indicate that the science was designed an informed choice, and potentially to work around an ob-
to teach computers to think like a human mind, rather stacle or a problem. Researchers argue that the brain is a
than developing an artificial system. This type of comput- type of machine, and can therefore potentially be repli-
ing integrates technology and biology in an attempt to re- cated. The development of neural networks was a large
engineer the brain, one of the most efficient and effective step in this direction.
computers on Earth. As the body of knowledge about the brain grows and sci-
However, with major advances in cognitive science, re- entists experiment more with cognitive computing, intelli-
searchers interested in computer intelligence became en- gent computers are the result. Smart computers which are
thused. Deeper biological understanding of how the brain capable of recognizing voice commands and acting upon
worked allowed scientists to build computer systems mod- them, for example, are used in many corporate phone sys-
eled after the mind, and most importantly to build a com- tems. Cognitive computing is also used in many navigation
puter that could integrate past experiences into its system. systems onboard aircraft and boats, and while these sys-
Cognitive computing was reborn, with researchers at the tems often cannot handle crises, they can operate the craft
turn of the 21st century developing computers which op- under normal conditions.

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
 Potential of Cognitive Computing and Cognitive Systems | 77

Figure 2: Projected Growth of Big Data (based on [1]).

At the turn of the 21st century, many researchers believed by new data, interactions and each other’s historical
that cognitive computing was the hope of a near future. data.
By replicating the human brain in computer form, re- – Meaning-based— performing natural language pro-
searchers hope to improve conditions for humans as well cessing and using embedded analytics to leverage
as gaining a deeper understanding of the biological reac- language structure, semantics and relationships.
tions that power the brain. Computers capable of reason – Highly interactive—providing tools and interac-
were beginning to emerge in the late 1990s, with hopes for tion designs to facilitate advanced communications
consciousness following. within the integrated system and incorporating
stateful human-computer interactions, data analy-
sis and visualizations.
2.3 Major characteristics of Cognitive Cognitive computing systems are able to handle ambigu-
Computing ity and a shifting set of variables. They can constantly re-
evaluate information based on changes in the user, task,
The major characteristics of cognitive computing systems context, goal or new information. They must understand
are [6]: the question or context before seeking answers. They may
– Information adept—being able to integrate big data offer multiple useful answers that are weighted for con-
from multiple heterogeneous sources and then syn- fidence or closeness to the query or topic. They turn big
thesizing ideas or answers from them. data into smart data and useful knowledge. Users are able
– Dynamic training and adaptive—learning and to interact with the system easily in a kind of continuing
changing as they receive new information, new "conversation." Like humans, cognitive computing sys-
analyses, new users, new interactions, new con- tems must be dynamic, and they must learn. Four layers
texts of inquiry or activity. of Cognitive Computing Systems can be identified, namely
– Probabilistic—discovering relevant patterns based – Static and dynamic Learning Systems
on context, statistically generating and evaluating – Data organization and interpretation
series of evidence-based hypotheses, predicting the – Architecture / Design of the system
probability of valuable connections, and returning – Core components
answers based on learning and deep inferencing.
This includes finding unexpected patterns—a kind
of machine-aided serendipity. 2.4 Cognitive Computing Building Blocks
– Highly integrated— has automated systems / work-
load management through which all modules con- To usher in a new era of cognitive computing, novel hard-
tribute to a central learning system and are affected ware, programming languages, applications and simula-

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
78 | Ahmed K. Noor

(a) of universities. Neuromorphic chips have the potential of

overcoming the physical limitations and considerably re-
ducing the power requirements of the traditional von Neu-
mann processors.
In 2011, a prototype neuromorphic chip containing 256
neurons and 262,144 synapses was developed by IBM. In
August 2014, IBM built a new chip with one million neu-
rons and 256 synapses. The new chip has 5.4 billion tran-
sistors and an-chip network of 4,096 cores. The chip con-
sumes only 70mW during real-time operation — orders of
magnitude less energy than traditional chips (Figure 3).
The goal is to simulate one trillion synapses using only 4
kW of energy [9].
The neurosynaptic cores are distributed and operate in
(b)
parallel. The cores integrate memory, computation, and
communication. Individual cores can fail and yet, like the
brain, the architecture can still function. Cores on the
same chip communicate with one another via an on-chip
event-driven network. Chips communicate via an inter-
chip interface leading to seamless scalability like the cor-
tex, enabling creation of scalable neuromorphic systems
(Figure 4). The neurosynaptic chip technology opens new
computing frontiers for distributed sensor and supercom-
puting, and robotic applications (Figure 5).

3 Cognitive / Smart Engineering

Figure 3: Neuromorphic Chips designed by Heidelberg group
Systems
and IBM. (a) The Heidelberg Chip features 384 neurons, 100,000
Synapses and operates at a speed of approximately 100,000 A Cognitive system is one that performs some of the func-
times biological real time; (b) The new IBM chip, called TrueNorth, tions of human cognition – learning, understanding, plan-
consists of 1 million programmable neurons and 256 million pro- ning, deciding, communicating, problem solving, analyz-
grammable synapses across 4096 individual neurosynaptic cores. ing, synthesizing, and judging. Some smart systems use
"brute force" computation to perform their tasks (like
tors are currently being developed. The new hardware in- some of the early concepts of driverless cars), others use
cludes new electronic neuromorphic machine technology machine (deep) learning to adapt to changing situations,
for processing sensory data, such as images and sound, detect novelty, seek out data, and augment human cogni-
and responding to changes in data in ways not specifi- tion. Emerging cognitive systems are being equipped with
cally programmed. This technology breaks path with the broad abilities in pattern recognition, natural language
traditional von-Neumann architecture used for the last 70 processing, complex communication, learning and other
years [7, 8]. domains that used to be exclusively human. They cover a
The DARPA’s SyNAPSE (Systems of Neuromorphic Adap- broad spectrum ranging from cognitive devices (e.g., neu-
tive Plastic Scalable Electronics) program initiated in 2008 rocam, and OrCam systems) to robotic machines and large
to model in silicon the massively parallel way the brain socio-technical systems (e.g., cognitive grids, cognitive in-
processes information as billions of neurons and trillions frastructures, and cognitive / smart cities). The cognitive
of synapses respond to sensory stimuli. The neurons also socio-technical systems are managed in a more holistic
change how they connect with each other in response to and intelligent way, using lean operational practices and
changing images, sounds or patterns. The program is be- cognitive technologies that can ultimately contribute to
ing undertaken by IBM, HRL labs, along with a number improving the reliability and responsiveness of customer
service and the whole economics of the systems. Cognitive

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
 Potential of Cognitive Computing and Cognitive Systems | 79

(a) systems will eventually impact every industry and every

enterprise. They will significantly increase human produc-
tivity through assisting, advising, and extending the capa-
bilities of humans.
Examples of some of the emerging cognitive engineering
systems are given subsequently.

3.1 Cognitive Materials

Increasing interest has been shown in the development

of cognitive material concepts through integrated sens-
ing and intelligence (sensorial material concepts), beyond
self-healing materials. The motivation for this work is
drawn from biological systems [10, 11]. The goal is to de-
velop a system that can inform engineers about how it is
feeling, where it hurts, how it changed its shape, and what
(b)
loads it is experiencing. Such materials are capable of real-
time self-monitoring, as well as using information on their
internal state to autonomously change major properties
(including thermal, electrical and acoustical properties).
The change in the internal state may not be predictable in
the design phase, since it may occur during the service his-
tory of the component.
Structures using cognitive material systems will have the
analog of a nervous system through the network of sensors
and communication facilities that link them.

3.2 Cognitive (smart) Cameras

Figure 4: The Brain Wall and Brain Cube. (a) The brain wall is a neu-
A cognitive camera can understand and interact with the
ral network visualization tool built by SyNAPSE project researchers
at IBM; (b) The brain cube will have 256 million neurons, and 64 surroundings, intelligently analyze complex scenes, and
million synapses. interact with the users. A prototype is being developed by a
multidisciplinary team from Penn State and Stanford uni-
versities (Figure 6). The project is inspired by biological vi-
sion, particularly the human visual cortex, and builds on
previous research at the University of Southern California,
and M.I.T. A major goal of the project is to build a low-
power cognitive devices that can replicate, and possibly
exceed, the human vision in a wearable form. The proto-
type will have embedded computation, and does not need
to be connected to the cloud (like Google glass).
Among the possible applications of cognitive cameras are
warning distracted drivers, when they have taken their
eyes off the road for a long period, and assisting the visu-
ally impaired.

Figure 5: Qualcomm Neurochip and a robot using the chip (Neuro

Bot).

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
80 | Ahmed K. Noor

Figure 6: Cognitive Camera developed by a research team led by

Penn State. The camera has an on-board computational power for
intelligent video processing.

3.3 Cognitive Robots

Significant work has been done in attempting to incor-

porate brain processing technologies into robots for bet- Figure 7: Myon the child-like cognitive robot developed by the
Neuro Robotics group at Humboldt university.
ter autonomy. The work was done under several titles, in
addition to cognitive robots, including neurobiologically-
inspired robots, neurorobots, and neuromorphic robots.
Much of the work is being funded by DARPA and NSF in
the US, the Human brain project in the European Union,
and the Australian Research council.
As the interaction and cooperation of robots with humans
increase, so does the demand for sophisticated robotic
capabilities associated with deliberation and high-level
cognitive functions. These include deep learning, on-time Figure 8: Collective Cognitive robots of the Artificial Life Labora-
decision, robustness and adaptability. Future cognitive tory of Graz, Austria. A swarm of autonomous underwater vehicles
robots will be equipped with advanced perception, dex- (AUVs) that are able to interact with each other and which can bal-
terity and manipulation to enable them to adapt to rea- ance tasks (interactions between/within swarms).
son, act and perceive in changing, incompletely known,
and unpredictable environments. These capabilities en-
operate in tasks. They could be used for biological moni-
able the robots to serve as effective human assistants or
toring, or for searching, maintaining, and even harvesting
companions [12, 13].
in underwater environments.
An early generation of cognitive robots is the human-like
The swarm will need the robustness and stability to func-
robot Myon, developed by the Neurorobotics research lab
tion under dynamically changing conditions. The vehicles
of Humboldt University in Germany (Figure 7). All the body
will interact with each other and exchange information,
parts of Myon can be completely removed during the op-
resulting in a cognitive system that is aware of its environ-
eration and flange-mounted again. The body parts retain
ment, of local individual goals and threats, and of global
their separate functionality because each one has its own
swarm-level goals and threats.
energy supply and computational ability. The neural net-
As shown by natural swimming fish swarms, such mech-
work is distributed over the decentralized robot.
anisms are flexible and scalable. The usage of cognition-
A European consortium, led by the University of Graz
generating algorithms can allow robots in the swarm to
and including both biological and technical institutions,
mimic each other’s behavior and to learn from each other
is creating a swarm of cognitive, autonomous underwa-
adequate reactions to environmental changes.
ter robots (Figure 8). The goal of the project is to develop
robotic vehicles that can interact with each other and co-

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
 Potential of Cognitive Computing and Cognitive Systems | 81

The plan includes investigating the emergence of arti- in driver assistance facilities, and driverless cars are de-
ficial collective pre-consciousness, which leads to self- scribed subsequently.
identification and further improvement of collective per-
– GM introduced driver assist package into their 2012
formance. In this way, several general principles of swarm-
Cadillac for detecting road hazards, and drawing the
level cognition will be explored to assess their importance
attention of the driver to them. The package uses
in real-world applications.
long-range radar that scans for objects up to 150 me-
The results can be exploited for improving the robustness,
ter away, short range radars, video and other cam-
flexibility, and efficiency of other technical applications in
eras, ultrasonic sensors, a central embedded com-
the field of information and computing technology.
puter for processing information and identifying ob-
In June 2014, the European Commission, along with 180
jects around the car.
companies and research organizations (under the um-
– Texas Instruments has introduced a family of ad-
brella of euRobotics), launched the world’s largest civil-
vanced driver assistance system devices (ADAS) that
ian research and innovation program in robotics. Cover-
can implement a number of driving features, includ-
ing manufacturing, agriculture, health, transport, civil se-
ing pedestrian detection, emergency braking, traf-
curity, and households, the initiative is called SPARC and
fic sign recognition, lane departure assistance, cross
aims at developing technologies including smart indus-
traffic alert adaptive cruise control, blind spot detec-
trial robots, autonomous cars, and drones.
tion, high-beam and park assist.
A cognitive / autonomous spherical robot concept has
been proposed by IBM. The robot has multi-modal sens- Information from multiple cameras (front-facing, side,
ing, including image and sound, and could be deployed in and rear), and various sensors are processed by the ADAS
a disaster area for search and rescue missions. An internal hardware simultaneously. The data then provides driver
mechanism would allow it to roll around an environment assistance for enhanced safety and driving. For example,
to survey areas and identify persons in need, the condition headlights can be aimed and adjusted dynamically based
of the zone and possible hazards. It could also communi- on external driving conditions.
cate with people it finds and guide them to safety through – Volvo began testing semiautonomous cars on city
speakers and a video display. streets. By 2017, a fleet of 100 cars will be part of
an experiment for autonomous handling of lane fol-
lowing, speed adaption, and merging traffic in the
3.4 Cognitive Cars Swedish city of Gothenberg.
– A number of companies, including Audi, Mercedes,
Cognitive cars are equipped with integrated sensors, cam- GM, and Nissan, are working on ways to automate
eras, GPS navigation system and radar devices that pro- the entire parking process. First , the car moves past
vide coordinates and information gathered on the road to a parking area, scanning for the marked edges of
other cars, equipped with the same car-to-car communica- parking slots and for obstacles (such as parked ve-
tion system. Automakers and their research partners are hicle). Different vantage points are provided by the
currently working toward creating cars that think on be- several cameras and ultrasound sensors in the cars.
half of drivers and passengers and act proactively, serving The information is processed to classify the parking
up infotainment and driver assistance features in antici- area according to its structure. Then the car moves to
pation of what’s needed or wanted. The basic elements in the next vantage point, and continues the process in
every infotainment or telematics UI are a touchscreen dis- order to offer a selection of possible parking slots.
play in the dashboard, voice recognition, control buttons – GM is working on crash-proof (crash-avoiding) car
or wheels on the steering wheel. concept. The concept relies on the use of active
The new technologies are intended to serve and protect safety systems, instead of the passive safety features
drivers and passengers, and ultimately render the human used in modern cars (including air bags, and frames
drivers superfluous. Driver assistance systems are laying that absorb the impact to protect the occupants dur-
the groundwork for autonomous (self-driving or driver- ing crash). The active safety systems use, in addition
less) cars. to the passive safety tools, visual alerts, loud sound,
The advanced technologies that make cognitive and self- and seat vibrations to alert the driver to imminent
driving cars have been filtering into commercial products danger (e.g., approaching cross traffic).
at a fast rate [14, 15]. Some of the recent developments

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
82 | Ahmed K. Noor

tool, which is slated for use in California starting in 2015,

can be fitted for nearly any vehicle. It includes two cam-
eras, a radar mechanism, GPS, inertial sensors and an on-
board computer, as well as actuators that control the car’s
steering, acceleration and braking actions. Using this soft-
ware/hardware combination, the Cruise RP-1 constantly
scans the road to keep the car operating within safe pa-
rameters in relation to other cars and the boundaries of the
driving environment.
Figure 9: Google’s two-seater self-driving car prototype eliminates
steering wheel, accelerator and brake pedals.

3.5 Cognitive Aircraft / Unmanned Aerial

– Nissan recently unveiled its neuro car concept- a car Vehicles (UAVs)
equipped with a Nismo smart watch that can moni-
tor a driver’s biometric information. Nippon Sangyo A paradigm shift is taking place in the design of
(Japan Industries) plans to monitor not just the car UAVs – from automated to autonomous and cogni-
but also the drivers’ vital signs, including brain- tive UAVs. Whereas automated UAVs have determinis-
wave activity and skin temperature. The company tic, or pre-determined behavior, and are controlled by
plans to use electrocardiograms (ECG) and elec- humans, cognitive UAVs make decisions that involve
troencephalograms (EEG) in the future to capture a non-deterministic, stochastic, and emergent behavior.
range of heart and brainwave data of the driver. They behave in ways that are not pre-planned and pre-
The watch and car will be in constant communication, the programmed.
car can update the driver via his watch whenever it needs In August 2014, BAE Systems has unveiled four futuristic
an oil change or its tires rotated. It will remind the driver technologies they believe could be incorporated in mili-
when it needs to be serviced. tary and civil aircraft in 2040 or even earlier. The four tech-
Since it is impossible to plan ahead for every single sce- nologies are autonomous UAVs, aircraft parts that can heal
nario that an autonomous car might have to handle, one themselves in minutes - called the Survivor, a new type of
of the key requirements of autonomous (self-driving) cars long-range aircraft which divides into a number of smaller
is to have cognitive capabilities (being able to learn and aircraft when it reaches its destination, dubbed the Trans-
make decisions on the fly). Recent self-driving car activ- former, and a directed energy weapon that could engage
ities include o Google self-driving cars, which in April missiles at the speed of light, destroy them, and protect
2014 have surpassed 700,000 autonomous accident-free the people on the ground.
miles. This was done by improving the software that can The Survivor is a lightweight adhesive fluid built inside the
detect hundreds of distinct objects simultaneously. The aircraft which allows jets to quickly heal themselves from
self-driving cars can ingest massive amounts of data in a damage sustained in flight. It is released to quickly ’set’ in
very short amount of time, explore multiple scenarios, and mid-flight and heal any damage.
eventually run simulations to insure that their decision are The Transformer is a flexible aircraft system that com-
as safe as possible. The cars pay attention to pedestrians, bines smaller jets for more efficient travel, before having
buses, stop signs, and a cyclist making gestures that indi- them split apart to quickly adapt to any scenario. BAE re-
cate a possible turn, in a way that a human driver cannot, searchers feel that by combining the jets on longer jour-
and they never get tired or distracted. The cars can deal ney there could be potential gains to be made in terms of
with changing environments and some level of dynamic increased range and fuel savings by cutting the amount
uncertainty. The fully autonomous, two-seater electric car of drag, scientists claim. All the four technologies are still
prototype, recently unveiled by Google, with the steering at the drawing-board stage, but BAE Systems is confident
wheel, accelerator and brake pedals eliminated. The cars about the prospects of them becoming a reality.
can go up to 25 mph. Google is building about 100 proto-
types of this sort and plans to conduct tests in versions that
retain the manual controls later this summer (Figure 9). 3.6 Cognitive Production Systems
o A new autopilot tool, Cruise RP-1, to enable hands free
driving on highways has recently been advertised. The Today’s manufacturing is facing many challenges, in-
cluding uncertainties of continuously and unprecedented

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
 Potential of Cognitive Computing and Cognitive Systems | 83

abrupt-changes in market demands, increasing number Cognitive machines could improve their performance and
of product variants, smaller lot sizes, enhancement of develop new capabilities through mining data and apply-
product quality, increasingly shorter time-to-market, and ing learning algorithms that are specific to their applica-
low cost production requirements. Manufacturing value tions. They could also exchange information with other
chains are distributed and dependent on complex infor- systems and provide updates to higher-level control sys-
mation and material flow requiring new approaches in- tem, helping to create more efficient and flexible factories.
side and outside the factory both on process and product Within all of this, the cognitive machines would remain
lifecycle levels. They have to respond faster and more effi- aware of their capacities, as well as their limitations. Cog-
ciently to higher complexity and frequently changing de- nitive manufacturing is characterized by capabilities and
signs. Cognitive production systems concepts, including visions for moving beyond "smart" manufacturing toward
cognitive machines and processes and cognitive factories, cognitive factories and systems that have the capacity
can be applied to address some of challenges [16]. This to monitor and evaluate manufacturing performance and
is accomplished through applying the principles of cogni- then propose process and operations improvements based
tion into every aspect of industrial production – from de- on using sensing technologies and multifaceted data for
sign to final assembly and quality control, and potentially real-time observation of human workers in industrial or-
to the end of the product life cycle. ganizations to increase the understanding of human cog-
The cognitive capability in production systems can be ac- nitive processes and transfer this knowledge to cognitive
complished through the development of cognitive reason- factories.
ing engines, or distributed intelligence agents, that are de- Cognitive factories are industrial autonomous systems that
ployed throughout the production system at three different introduce artificial cognitive capabilities to the dynamic
hierarchical levels: a) the manufacturing process level, b) allocation of production services, and the control of pro-
the manufacturing system level, or factory level, and c) the duction systems in order to overcome today’s bound-
production system logistical level, or supply chain level. aries [17]. These include autonomous machining, auto-
Actions must include validation/demonstration elements mated programming of industrial robots, human –robot-
and involve stakeholders covering the whole value chain. cooperation, knowledge –based quality assurance and
In addition, mobile transportation robots are used to en- process control. The use of cognitive factories has the po-
able flexible routing – A team of robots can produce new tential of enhancing the adaptivity of production systems,
products from a number of semi-finished products. as well as the quality of products.
Researchers at Harvard and MIT have recently build a
robot that assembles itself into a complex shape in four
minutes flat, and crawls away without any human in- 3.7 Cognitive Wireless Technologies
tervention. The advance, demonstrates the potential to
quickly and cheaply build sophisticated machines that in- Cognitive (or smart) radio networks like xMax system of
teract with the environment, and to automate much of the xG technology, Inc. represents an innovative approach to
design and assembly process. The method draws inspira- wireless engineering in which radios are designed with
tion from self-assembly in nature, such as the way linear an unprecedented level of intelligence and agility. It en-
sequences of amino acids fold into complex proteins with ables more efficient use of scarce expensive wireless spec-
sophisticated functions. trum though using a number of technologies, including
Among the possible applications of this concept is that of dynamic spectrum access (DSA), interference mitigation,
a swarm of robotic satellites sandwiched together so that full mobile handoff and software defined radio (SDR) ca-
they could be sent up to space and then assemble them- pabilities [18].
selves remotely once they get there-they could take im- Cognitive radios have the ability to monitor, sense, and
ages, collect data, and more. The concept is complemen- detect the conditions of their operating environment, and
tary to 3D printing, which also holds great promise for dynamically reconfigure their own characteristics to best
quickly and inexpensively manufacturing robotic compo- match those conditions [19].
nents. The longer term could be changing the way we think They automatically detect available channels in wireless
about manufacturing in that the machines fabricate them- spectrum, then accordingly changes its transmission or
selves. For example, cognitive 3D printers could receive reception parameters to allow more concurrent wireless
data and automatically design and build bespoke products communications in a given spectrum band at one location.
to meet individual demands. They provide a smarter, faster, and more efficient way
to transmit information to and from fixed, mobile, other

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
84 | Ahmed K. Noor

wireless communication devices. They have significantly hibit errors that can be automatically corrected and repro-
improved the wireless communication performances, in duced so that the pilot receives the correct readings and
a cost effective way, by exploiting underutilized licensed can adjust accordingly.
bands In addition to its benefits in making flight systems safer,
smart analytics could also be helpful in other applications
that rely on sensors, such as healthcare. Analyzing the pat-
3.8 Cognitive Flight systems, Airspace terns of data collected from sensors attached to the human
Systems, and other infrastructures body could detect, ahead of time, early signs of seizures or
heart attacks.
Cognitive Flight and aerospace systems are examples of
cognitive socio-technical systems, which attempt to am-
plify the human capability in performing cognitive work 3.8.2 Cognitive Airspace System
through integrating the technical functions with the hu-
man cognitive processes [20, 21]. Other examples include The airspace is a distributed and heterogeneous system
healthcare systems, energy, transportation and other large comprising diverse human and technological functions.
scale infrastructures. Enhancing the effectiveness of air traffic control and man-
The issues addressed in socio-technical systems include agement, and extension of the airspace system to handle
decision making in complex and dynamic information en- UAVs, requires equipping the facilities and resources in the
vironments, distributed collaboration, and management system with cognitive capabilities.
of extensively networked systems. For example, cogni- Modernization of air traffic management is already under
tive energy grids use variety of technologies in conjunc- way. The new systems envisioned in Europe and the U.S.
tion with big data to anticipate and smoothly meet the would begin automating a pilot’s tasks even before sub-
rapidly changing needs of different load centers dis- mission of a flight plan. From a hotel room, hours before
tributed around the grid. They intelligently source and al- a flight, the pilot could plug an electronic flight planner
locate energy from renewable sources (like solar, wind and into a tablet computer, automatically linking to the plane’s
hydro), and non-renewable sources, as required to ensure flight management system and the FAA air traffic network
smooth and reliable delivery. center. He could choose one of several routes, from the
fastest to the least expensive. He would receive weather
information, including the wind’s speed and direction.
3.8.1 Cognitive / smart data for flight systems Within seconds, an optimal route would be calculated and
a touchdown time determined, accurate to within two sec-
Among the recent activities on cognitive technologies for onds. Two out of three flights would be flown automati-
socio-technical systems is the work on "cognitive / smart cally: A plane’s own systems, linked to the airline’s opera-
data" technologies by researchers at the Rensselaer Poly- tions and FAA computers, would operate the aircraft, from
technic Institute. The work is funded by the Air Force Of- the closing to the opening of the passenger cabin doors.
fice of Scientific Research, and aims at turning passive Schedule disruptions caused by late passengers or sudden
data systems into active ones by enabling them to search changes in weather would be managed by the networked
for patterns and relationships, discover knowledge in data computer systems, with the pilot alerted to the changes.
streams, as well as incorrect data generated by faulty sen- Very rarely would a pilot have to intervene during an auto-
sors, or other hardware failures, such as those that con- matic flight, and usually only because the network offers
tributed to the Air France 447 crash in June 1st, 2009. Dur- faster or more fuel-efficient routing.
ing that flight, important sensors failed, and reported er- It is somewhat clear what the key enabling technologies
roneous airspeed data. But the autopilot didn’t know that, and procedures will be, how much they will cost to intro-
and it acted as if the data were correct. duce and what benefits they will bring. However, what is
The smart data system uses mathematical and program- less clear is how the global network will be able to operate
ming elements that searches for patterns and relationships when parts of it are degraded, how soon all aircraft oper-
that indicate hardware failure. Active data has been in- ators will adopt compliant technologies, how controllers
corporated into a software system called the "Program- will accept their changing roles, and how a global net-
ming Language for spatiO-Temporal data Streaming ap- centric traffic management system will be regulated and
plications," or "PILOTS," which treats air speed, ground certified.
speed and wind speed as data streams that sometimes ex-

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
 Potential of Cognitive Computing and Cognitive Systems | 85

Although significant technological advances have been Recently, the creators of Siri, have announced a smart as-
made by the Unmanned Aerial System (UAS) community, sistant successor Viv (the global brain), that is currently
critical research is needed to fully understand the impact being developed and which can recognize the user’s per-
of UAS operations in the National Air Space (NAS). There sonal preferences and a near-infinite web of connections
has also been little research to support the equipment de- to answer almost any query and perform almost any func-
sign necessary for UAS airworthiness certification. In the tion. Google is working on a more ambitious cognitive as-
near- to mid-term, UAS research will need to focus on tech- sistant called the "Google Brain" that melds computer sci-
nology deemed necessary for UAS access to the NAS. ence with neuroscience.
Emerging cognitive assistants are proactive autonomous
agent systems (Cogs), a class of more advanced software
programs, which are designed to follow and interact with
4 Emerging Cognitive technologies people (and other cogs and services) inside and across cog-
and Environments nitive environments. Cogs use cognitive analytics to pro-
cess natural language and do pattern recognition.
4.1 Cognitive Interfaces

The next phase of user interface architecture is multisen- 4.3 Cognitive Environment
sory and multimodal, so the machine is capable of recog-
Research on cognitive environments is moving from over-
nizing a gesture while interpreting the user’s voice simul-
coming physical, connectivity limitations to overcom-
taneously. Some of the recent work is devoted to creating
ing productivity and complexity limitations through en-
machines that has pattern and object recognition capabil-
hanced cognition [22]. Among the recent developments
ities, and can think on behalf of their users and act proac-
are integrated sensor system for environmental and user
tively, serving as user assistance features in anticipation of
activity monitoring, wearable technologies (including,
what’s needed or wanted.
annotated-reality glasses), advanced telepresence and vi-
Future user interfaces and human machine interface sys-
sualization facilities, tools for reducing cognitive load, and
tems are likely to be deceptively simple, yet so pioneering
Symbiotic cognitive systems incorporating Cogs.
they could easily pass for special effects in a science fic-
Future cognitive environments will include learning and
tion film—including neural interfaces, steering wheel sen-
training facilities which can predict performance and
sors in cars capable of reading a driver’s mind and touch-
learning needs, and distributed healthcare networks that
screens that morph into physical buttons and switches.
could draw on vast quantities of medical data to help
Cognitive interfaces will allow new decision-making expe-
physicians in providing the right treatment to patients
riences, such as making a critical maintenance decision on
with unusual conditions.
a gas pipeline. The power of these decision spaces is that
they give users real insights, not just information; by pro-
viding valuable insight, users are one step closer to taking
action.
4.4 Cognitive enterprise
Future enterprises will have cognitive environments, in-
corporating variety of cognitive assistants and cognitive
4.2 Cognitive Assistants (Cogs) tools and facilities. Their workforce will be engaged in
complex applications requiring in-depth data analysis and
Since the beginning of the 21st century, a number of AI ac- adaptability. The algorithms used in these applications
tivities were devoted to developing intelligent personal as- will be very hard to design and require processing exten-
sistants (beyond Question / Answering systems). Among sive and complex big data sets (using big data predictive
these activities is the large AI project funded by DARPA- and prescriptive analytics).
the Personalized Assistant that Learns (PAL) program.
The Cognitive Assistant that Learns and Organizes (CALO)
project led by SRI was part of this program. A number of
personal assistants on mobile devices resulted from CALO, 5 A look at the future
including Apple’s Siri, Google Now, Dragon Mobile Assis-
tant, Nina Mobile, and Microsoft Cortana. Cognitive computing is a transformative area of comput-
ing, and a major driver for knowledge automation work,

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
86 | Ahmed K. Noor

Figure 10: Augmenting and Amplifying human capability through cognitive computing and cognitive technologies.

both as an emerging enterprise disruptor, and future cog- able symbiotic interaction with the environment, drive in-
nitive enterprise enabler. The confluence of Cognitive tech- novation, flexibility, and cost reduction.
nologies with technologies such as cloud, mobile, wear- The new products will have higher level of intelligence
able devices, Internet of Things (IOT), big data and vi- than current smart products, and include cognitive cyber-
sual analytics will amplify their impact. A new genera- physical systems, with embedded intelligence, mecha-
tion of cognitive devices, facilities and systems will be de- tronic and adaptronic components, that can monitor their
veloped. The coming years will witness the infusion of own state, and are able to self-configure, self-optimize,
cognitive technologies, devices, tools, facilities into de- self-protect, and self-heal. They will be able to communi-
cision support systems, engineering processes, practices cate with other cognitive products.
and systems. Humans and machines will be working to-
gether to amplify human capabilities, particularly those
associated with insight, finding relevant patterns in dy-
namic big data, making optimal choices, and discovery of
6 Concluding remarks
new generation of cognitive products ([23, 24], and Fig-
In this paper a brief review is given of cognitive computing
ure 10). A radical shift in engineering software and en-
and some of the cognitive engineering systems activities.
gineering practice will occur. Engineers will use generic
The potential of cognitive technologies is outlined, along
cognitive assistants (cogs) to build customized cogs for
with a brief description of future cognitive environments,
various applications (such as modeling, simulation, and
incorporating cognitive assistants - specialized proactive
knowledge capture and representation in product design).
intelligent software agents designed to follow and inter-
The field of Cognitive product engineering will be de-
act with humans and other cognitive assistants across the
veloped as a new procedural paradigm, and a frame-
environments. The cognitive assistants engage, individu-
work, for describing, designing, manufacturing, operat-
ally or collectively, with humans through a combination
ing, and servicing cognitive products [25]. It will incor-
of adaptive multimodal interfaces, and advanced visual-
porate deep learning, predictive and prescriptive analyt-
ization and navigation techniques.
ics, emergent engineering concepts, adaptive multimodal
Cognitive computing, technologies and systems are a pas-
human-machine interfaces, along with other visuo-spatial
sage through a world yet to be imagined, and future de-
cognition, computational and collaboration tools to en-
velopments in these areas will continue to push the limit
of what is possible. Planned and future activities include

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
 Potential of Cognitive Computing and Cognitive Systems | 87

neuromorphic chips that monitor the performance of en- the-hype-97685.aspx.

gineering systems (using data generated by variety of sen- [7] Hof R. D., Neuromorphic chips, MIT Technology review,
117, 2014, 54-57, http://readingpack.com/article/?i=
sors) and provide early warnings of potential problems,
617f509ed7e30954522a5fe846c379b00cdc7ac6.
and suggest possible remedial actions; cognitive systems
[8] Hasler J., Marr B., Finding a roadmap to achieve large neuromor-
which use image and speech recognition as their eyes phic hardware systems, Frontiers in Neuroscience, 2013, 7, 1-
and ears to recognize patterns, understand their environ- 29, http://journal.frontiersin.org/Journal/10.3389/fnins.2013.
ment, and interact more seamlessly with humans; cogni- 00118/full.
tive products that are capable of knowing their operational [9] Greenemeier L., Brain-Inspired computing reaches a new
milestone, Scientific American blogs, 7 August 2014,
environments and can act autonomously; and smart cities
http://blogs.scientificamerican.com/observations/2014/
that can optimize their infrastructures, networks and other 08/07/brain-inspired-computing-reaches-a-new-milestone/.
facilities. [10] Frommberger L., Freska C., From Smart materials to cognitive
A holistic perspective and a comprehensive strategy are materials – requirements and challenges, Proceedings 10th
needed to develop the discipline of cognitive engineering Biannual Meeting of the German Society for Cognitive Sci-
ence (KogWis 2010), Eds. Johannes, H. et al., Universitätsverlag
and to put its various activities on an ambitious trajec-
Potsdam, 2010, 103-104, http://cosy.informatik.uni-bremen.
tory that pushes the frontiers of innovation, discovery, and
de/sites/default/files/u11/CognitiveMaterials.pdf.
economic development. A step toward implementation of [11] Lehmhus D., Lang W., Bosse S., Material-Integrated Sensing
that strategy is the development of a cognitive innova- and Intelligent Systems. A review, EPoSS Annual Forum, Cork,
tion ecosystem for the engineering workforce. The contin- September 2013. http://www.smart-systems-integration.
uously expanding major components of the ecosystem in- org/public/documents/presentations/presentations-at-
the-joint-event-mnbs-2013-and-eposs-general-assembly-
clude integrated knowledge discovery and exploitation fa-
annual-forum-2013/eposs-annual-forum-2013-26-september-
cilities (incorporating predictive and prescriptive big data 2013/3-2_Dirk%20Lehmhus_ISIS_Material-Integrated%
analytics); novel cognitive modeling and visual simula- 20Sensing%20and%20Intelligent%20Systems.pdf.
tion facilities; cognitive multimodal interfaces; and cogni- [12] Ishiguro H., Osaka M., Fujikado T., Asada M. (eds.),
tive mobile and wearable devices. The ecosystem will pro- Cognitive neuroscience robotics: synthetic approaches
to human understanding, Springer-VerlagGmbH, 2014,
vide timely, engaging, personalized / collaborative, learn-
http://cosy.informatik.uni-bremen.de/sites/default/files/
ing and effective decision making. It will stimulate creativ-
u11/CognitiveMaterials.pdf.
ity and innovation, and prepare the participants to work [13] Domenico P., Future robots: towards a robotic science of
in future cognitive enterprises and develop new cognitive human beings, John Benjamins Publishing Co., Amsterdam,
products of increasing complexity. 2014, http://books.google.com/books?id=DDTOAwAAQBAJ&
pg=PA482&dq=cognitive+robotics+2014&hl=en&sa=X&
ei=J2PzU8DGGdHfsATpwYCIBw&ved=0CDAQ6AEwAA#v=
onepage&q=cognitive%20robotics%202014&f=false.
References [14] Gadsden S. A., Habibi S. R., The future of automobiles: cogni-
tive cars, Proceedings of 22nd Canadian Congress of Applied
Mechanics (CANCAM), Halifax, Nova Scotia, 2009, 111-112,
[1] Kelly J. , A new era of computing, IBM Investor Briefing,
http://www.coreng.ca/gadsden_publications/gadsden_conf_
2012, http://www.ibm.com/investor/events/investor0512/
005.pdf.
presentation/09_Research.pdf.
[15] Calem R. E., Cognitive cars, i3 it is innovation, 18 Septem-
[2] Kelly J. E., Hamm S., Smart machines: IBM’s Watson
ber 2013, http://www.ce.org/i3/Features/2013/September-
and the era of cognitive computing, Columbia University
October/Cognitive-Cars.aspx.
Press, New York, 2013, http://www.amazon.com/Smart-
[16] Bannat A. et.al., Artificial cognition in production systems,
Machines-Cognitive-Computing-Publishing/dp/023116856X/
IEEE Transactions on automation science and engineering,
ref=sr_1_1?s=books&ie=UTF8&qid=1393450316&sr=1-
8, January 2011, 148-174, http://www6.in.tum.de/Main/
1&keywords=cognitive+computing.
Publications/Bannat2010a.pdf.
[3] Gill D., The new era of cognitive computing, https://www.
[17] Zaeh M. F. et al, The cognitive factory, in Changeable and Recon-
youtube.com/watch?v=KiihVEcj-lY.
figurable Manufacturing Systems, Elmaraghy H. (ed), Springer
[4] Wladawsky-Berger I., The era of cognitive computing, 1 July
Series in Advanced Manufacturing, 2009, 355-371, http://link.
2013, http://blog.irvingwb.com/blog/2013/07/the-dawn-of-a-
springer.com/chapter/10.1007%2F978-1-84882-067-8_20.
new-era-in-computing.html.
[18] Jefferies N., Next generation wireless for a cognitive and energy
[5] Modha D. S. et al., Cognitive computing, Communications of
eflcient future, presented at the ETSI future mobile summit,
ACM, 54, August 2011, 62-71, http://dl.acm.org/citation.cfm?
Mandelieu, France, 21 November 2013, http://docbox.etsi.org/
id=1978559&bnc=1.
Workshop/2013/201311_FUTUREMOBILESUMMIT/05_WWRF_
[6] Reynolds H., Feldman S., Cognitive computing: Beyond the
JEFFERIES.pdf.
hype, KM World, 27 June 2014, http://www.kmworld.com/
[19] Meghanathan N., Reddy Y. B., Cognitive radio technology
Articles/News/News-Analysis/Cognitive-computing-Beyond-
applications for wireless and mobile Ad Hoc networks, IGI

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access
88 | Ahmed K. Noor

global, 2013, http://www.igi-global.com/book/cognitive- [23] Metzler T., Shea K., Cognitive products: Definition and frame-
radio-technology-applications-wireless/74190#table-of- work, International Design Conference – Design 2010 (17-
contents. 20 May 2010, Dubrovnik, Croatia), 2010, 865-874, http://
[20] Haque J. et al, Cognitive aeronautical communication system, mediatum.ub.tum.de/doc/1167283/1167283.pdf
International Journal of Interdisciplinary Telecommunications [24] Abramovici M., Stark R. (Eds.), Smart Product Engineering, Pro-
and Networking, 3, January-March 2011, 20-35, http://www. ceedings of the 23rd CIRP Design conference, Lecture Notes
irma-international.org/viewtitle/52982/. in Production Engineering, Bochum, Germany, 11-13 March
[21] Lintern G., The Airspace as a cognitive system, The inter- 2013, Springer, Berlin, 2013, http://link.springer.com/book/
national journal of aviation psychology, Taylor and Francis 10.1007%2F978-3-642-30817-8.
group, LLC, 2011, DOI: 10.1080/10508414.2011.537556, [25] Lee J.D., Kirlik A. (Eds.), The Oxford handbook of cognitive
http://www.cognitivesystemsdesign.net/Papers/Lintern% engineering, Oxford University Press, New York, 2013, http:
20(2011)%20The%20Airspace%20as%20a%20Cognitive% //global.oup.com/academic/product/the-oxford-handbook-
20System.pdf. of-cognitive-engineering-9780199757183;jsessionid=
[22] Goel A., Swaroop V., Wiltgen B., Helms M., Cognitive, collab- D4E6F752E381F1A74F36826D6C412CDB?cc=us&lang=en&#.
orative, conceptual and creative – four characteristics of the
next generation of knowledge-based CAD systems: A study
in biologically inspired design, Computer-Aided Design, 2012,
44, 879-900, http://www.sciencedirect.com/science/article/
pii/S0010448511000820

- 10.1515/eng-2015-0008
Downloaded from PubFactory at 08/23/2016 10:40:28PM
via free access