The Construction of an Indoor Floor Plan Using a

Smartphone for Future Usage of Blind Indoor

Navigation.
J. A.D.C.Anuradha Jayakody Iain Murray
Department of Electrical and Computer Engineering Department of Electrical and Computer Engineering
Curtin University Curtin University
Perth,Western Australia Perth,Western Australia
[email protected] [email protected]

Abstract—Most blind people require assistance to navigate within introduced the Industry Foundation Classes (IFC) [2].
unfamiliar environments as there is often insufficient information OpenBIM [4] describes the database that contains information
about the buildings available to them. To address aspects of this about IFC compatible applications for the whole range of
problem, this paper describes the “AccessBIM” model as an design and construction disciplines, including architectural,
approach to facility management in which a digital
structural, building services, facility management and model
representation of the indoor building features is used to facilitate
the exchange and interoperability of real-time information in viewing. To manage and centralize the construction based on
digital format which can assist blind people to independently the open standard IFC. This research proposes to build an
access unfamiliar building indoor environments. The model Accessible Building Information Model (AccessBIM) to
driven architecture that can be implemented for way finding and centralise and manage the information of the built
data synchronization, generating, in real-time topological map, to environment.
assist vision impaired individuals to navigate known and
unknown indoor environment. II. RELATED WORKS

Keywords-component; Assistive Technology; vision

impairment; indoor navigation; SmartSLAM; FastSLAM; This work is a part of a larger project that is developing an
IndoorOSM; crowdsourced database; AccessBIM indoor navigation system for vision impaired people using
mobile devices such as smart phones and tablets. The
I. INTRODUCTION following sections describe the literature review and the
The number of blind and low vision people is increasing by conceptual model that authors proposed.
1 million to 2 million per year. This estimate indicates that the Over the past two decades, significant research has been
total number of blind people will reach 100 million in 2020, dedicated to navigation assistance for the blind or visually
not including low vision people [1]. The problems become impaired persons. Many of these navigation devices can be
more severe when it comes to going from one place to another categorized into basic obstacle avoidance systems, for
independently. Society has come to an age of supermarkets, example, the NavBelt from Shoval et al [12] that produces a
shopping malls, office complexes, huge warehouses etc. In 120 degree view ahead of the users’ current location. This
such complicated surroundings people with vision information is translated into a stereophonic acoustical sound
impairments may feel overwhelmed. While navigation that allows the user to notice if a certain direction is blocked.
systems for outdoor environments are readily available, Similar to this approach the Haptica Corporation developed
navigation within buildings still poses a challenge. The Global Guido© [7] a robotic walking frame equipped with a sonar
Positioning System (GPS), the dominant system offering sensor. According to Kulyukin et al promising approaches rely
location information outdoors, suffers a poor indoor on the usage of robotic assistances [17]. Although they work
performance due to low signal availability, as GPS signals are very well in the new, yet unknown indoor environments they
not designed to penetrate through most construction materials. still rely on certain provided infrastructure. For outdoor
Many indoor positioning techniques have been developed, navigation several systems have been proposed [8, 9, 11, 14]
most of which rely on fixed positions to determine the location or are commercially available [16]. Although they share
of tagged devices [2]. Traditionally the buildings are common characteristics with indoor navigation systems, this
documented using building maps and plans stored in paper will not go into further detail regarding these systems, as
electronic form with tools such as computer-aided design they often rely on technologies that are not available in indoor
(CAD) applications. Storing the maps in an electronic form is environments - for example GPS for tracking.
already pervasive but CAD drawings are not adequate for the
requirements of effective building models aimed at indoor Sonnenblick [13] evaluated and implemented a large-scale
navigation systems. To reach a higher level of data integrity indoor navigation system for blind individuals at the
and utilization potential of building information models, the Jerusalem Center for Multi-Handicapped Blind Children. It
International Alliance for Interoperability (IAI) [2] has relies on specially installed infrared beacons for orienting

978-1-4799-6629-5/14/$31.00 2014
c IEEE 145
blind users and on a custom-built end-user device. However,
the “Chatty Environment” by Coroama [6] is focusing on
technology that does not require special hardware, but rather B. Processing Stage 2
equipment that is already generally available. This system uses
standard 802.11 Wi-Fi for positioning and PDAs as end-user
devices, which increases the chances for a widespread use. In At the second stage, two sub processes (Sub Process
a later work they even extend this system to special routing 1 & Sub Process 2) are executed, Sub process 1 includes
algorithms focused to the needs of the blind [15]. However, smoothing, vertex constriction and sphere contraction and sub
their system does not really address common problems with process 2 contain as modification of a relaxation algorithm for
Wi-Fi positioning and its rather coarse precision. error correction of the data (Fig. 2). Using a relaxation
algorithm as a first step can calculate the estimated position
and its variation proposed, the map creation and the post-
III. NAVIGATION MODEL editing put every vertex to its expected position [6].

Navigation in an unfamiliar indoor environment is a Sub Process 2

challenge for vision impaired individuals. This section
explains the methodology to identify indoor environments
using spatial information with an indoor environment using
“AccessBIM [13]” to assist blind and vision impaired
individuals to navigate inside an unknown building
environment. To address the issues involved in the application
of collaborative indoor mapping, this paper discusses four
processing stages Sub Process 1
A. Processing Stage 1(data pre-processing)
In the data pre-processing stage, gait analysis [20] and the
FootPath algorithm [22] is to be utilized as shown in Fig.1 In Fig. 2: Block diagram of the Second Stage
navigation, pedestrian dead reckoning (PDR) is the process of C. Processing Stage 3
calculating one's current position by using a previously
determined position, or fix, and advancing that position based In the third stage a spatial database is made available
upon known or estimated speeds over elapsed time and course to store data. The database includes a classic navigation graph,
[12]. By recording direction and distance of movement, one additional vector data of probable indoor movement traces,
can estimate the current position by adding these motion which would allow for generation of navigation directions as
vectors to a given start position. The human Gait (IMU) well as optimal cartographic information of calculated routes.
analysis and image processing can use different sensor data, Objects are classified into two main categories as “unmovable
such as an accelerometer and an electronic compass sensor obstruction”, and “movable obstruction” within the
integrated into a smartphone [20, 21]. By tracking a user's environment; a movable obstruction-object can be moved here
path, PDR approaches estimate the location of a user by and there within the indoor environment and “pathway”- the
knowing start position, stride length and the number of his paths along which they walk. The collected data should store
walked steps with their corresponding heading. the database inside the mobile phone as well as the cloud
environment to share the constructed maps. This stage
According to Paul Anthony Smith it is possible to estimate proposed study the possibility of using a simultaneous
the step length of a user by using the correlation between step localization and mapping (SLAM) techniques to generate
length “l” and body height “h”. The corresponding formula is maps. The SLAM problem is an active research field in
“l = h * 0:415” [4, 5]. Considering that the calculated step robotics. The SLAM describes how a mobile robot can
length is only a raw estimation of the travelled distance, the determine its own position without having localization
author proposed to develop an approach to fix occurring information, like a map, about its environment [1, 5].
errors, such as inaccurate compass readings, wrong estimated Therefore, the robot explores the surrounding area, starts to
distance and wrong step detection, during the navigation. create a map of this area and while it is creating the maps for
building with an existing Wi-Fi infrastructure which is
available inside the building and the robot tries to determine
its position created by using the map. The SLAM problem
appears in different domains, e.g., it can differentiate between
outdoor SLAM and indoor SLAM. In addition, there are
diverse strategies for solving the SLAM problem. Smart
SLAM provides the automatic creation and updating of indoor
plans. The main idea of SmartSLAM is to create an indoor
map, which consists of Wi-Fi fingerprints. For that,
Fig. 1. Block diagram of the data pre-processing stage

146 2014 International Conference on Contemporary Computing and Informatics (IC3I)

SmartSLAM uses Wi-Fi access points, which are installed an image, positioning it at a certain location, rescale and resize
within buildings, as landmarks. Additionally, it tracks a user's it and store a configuration file for the adjusted image. This
path by using an accelerometer sensor to detect steps and an configuration file allows us to load the image again later.
electronic compass sensor to detect directions. SmartSLAM
implements the typical components of SLAM. In addition, the
observation model consists of recording signal strengths with
their corresponding access points. Finally, the created map
consists of the recorded Wi-Fi access points, which the author
“Shin et al” proposed to use as landmarks [1, 2, 3, 4, 5,].

In the sub processes “OpenStreetMap” may be used

Fig. 3. The block diagram of the Fourth Stage
to create maps. The pre-mapping phase is important to collect
required parameters for the map creation, from the gait With the combination of the above mentioned four stages
analysis data [20] which is necessary to derive the estimated this model for mapping indoor environments can be accessible
step length of the user during the mapping process, and the for vision impaired mobile user to navigate indoor
map's name. Additionally, in this stage data from image environments that will be built using the given data. The
processing [21] will be incorporated, e.g., information of a model will provide more detailed information about buildings
door or other necessary parameter is the start position for map and their inner structures through shared a cloud in a real time
creation. The map generation process for indoor navigation in multi user environment.
this prototype involves both manual and automated work. To generate the maps in an unknown environment, it is
Each and every created map and every floor of the map a required to have displacement and direction as well as an
bounding box file is created. This is a configuration file that absolute position point of interest (POI). The user trail (T) (Fig
contains all the important parameters for a created map. It 4.) is uploaded to the mobile server over a WiFi/3G
should contain general information about the floor level of the connection. As given data sensor data make the user location
bounding file, whether plan of the floor is available, and it and trail drift from the true path. However, two opportunities
holds the world coordinates of the boundaries of the screen. are exploited to correct location and drift: (i) encounters with
Furthermore, it additionally should contain the storage the beacon (Beacon frame is one of the management frames in
location of the available map, the corners of the adjusted IEEE 802.11 based WLANs. It contains all the information
image file in world coordinates and the transformation matrix, about the network), and (ii) encounters with users who passed
which can be applied to the map. In the newly proposed the beacon recently. Both these events, if detected, provide a
architecture, sensor data will be collected from the external better estimate of the user's location [1].
environment for the purposes of map generation. Two sources
can be identified as the external sensor inputs, namely, sensor
data through image processing and sensor data through gait
analysis. Sensor data obtained from both the sources will be
directed to the specially designed API for processing. After
the completion of the error correction process, a decision
making process is initiated.

D. Processing Stage 4

The final stage of this methodology provides rapid

convergence of objects within the AccessBIM to assist real
time processing with multiple indoor users. The Java
OpenStreetMap Editor (JOSM) [10] is a Java-based editor for Fig. 4. Overview of AccessBIM Sytem,Received Gait analysis and the Image
creating and editing “OpenStreetMaps”. This application will process information as well as user encounters.The Server from user trails.
extend it by creating or modifying different plugins. First,
JOSM offers the functionality to download directly as well as The difference (in displacement and direction) between the
upload created maps to the OpenStreetMap servers. new/better location estimate and the old sensor-computed
Furthermore, JOSM allows exporting and importing location presents information about the nature of the drift [1].
OpenStreetMap maps into an XML formatted file. Overall, AccessBIM model uses this distance information to correct
JOSM gives the possibility to import created maps, post-edit both the current l user occasion and her (past) trail. Correcting
them by using the integrated functionalities and upload the the user trail is important to ensure more accurate routes
edited maps to the OpenStreetMap server. The JOSM has only between users. User trail is important to ensure more accurate
limited indoor support. Creating maps and adding all routes between users.
necessary indoor attributes, like floor levels, is time intensive.
The PicLayer [11] is a JOSM plugin that allows the loading of

2014 International Conference on Contemporary Computing and Informatics (IC3I) 147

 The IndoorOSM model can be described as follows,a building
is considered as a hierarchically structured object which
IV. NAVIGATION ALGORITHAM consists of several levels/floors. These floors are furthermore
divided into elements such as rooms, corridors, stairways,
The described proposed navigation algorithms consist two doors or windows. Within OSM, a building is therefore
main parts, as listed follows. mapped as a relation (main-relation). All the different floors
are also mapped as relations (level-relation), whereby these
• Data acquisition and smoothing algorithm for AccessBIM are considered as relation members (children) of the main-
[20]. relation. Furthermore, the different building parts (rooms, etc.)
• Map generation and error correction algorithm for are also mapped as relation members of the corresponding
AccessBIM [21]. level-relation. The IndoorOSM proposal aims at providing
detailed floor plan information, thus all building parts are
The “AccessBIM” model as an approach in which a digital typically mapped as closed ways, representing the polygonal
representation of the indoor building features is used to footprint of the corresponding building part.
facilitate the exchange and interoperability of real-time
information in digital format which can assist blind people to
independently access unfamiliar building indoor environments
[13].

In the newly proposed architecture (Fig. 5), sensor data

will be collected from the external environment for the
purposes of map generation. Fig. 5 two sources can be
identified as the external sensor inputs, namely, sensor data
through image processing and sensor data through gate
analysis. Sensor data obtained from both the sources will be
directed to the specially designed API for processing. Inside
the API a special function named dataSegment() will be
invoked. Based on a set of predefined rules dataSegment()
function will attempt to segment the incoming sensor data into
two categories namely, temporal data and landmark based data
semantics. The function will be called in an iterative manner
until all the sensor data is segmented and stored in the
corresponding tables within the Data Base (DB). During the
next stage of the proposed algorithm correctly segmented data
will be retrieved from the appropriate tables for the purpose of
decision making. Since the algorithm has to deal with
inaccuracies in the map creation, such as inaccurate sensor
data, wrong counted steps and an estimated stride length. All
these factors lead to errors resulting in wrongly created maps.
To deal with these errors, the authors propose to use the
relaxation algorithm during the error correction step. The
Fig. 5. Map generation and error correction algorithm for AccessBIM
tested algorithm was developed by Duckett “et al” [6]. With
this algorithm, the authors plan to correct inaccurate step
lengths during the map creation and also plan to use this V. TESTING LOCATION AND TEST RESULTS
algorithm during the post-editing to put every vertex to its
expected position. Considering that it needs to deal with Through the API, data that have been collected to the
inaccurate compass readings, hence proposed to modify the AccesBIM database, the ambition is to test the expected
algorithm. After the completion of the error correction process outcomes, so the collected data and obtain results an indoor
the decision making process comes to the role. SLAM will be environment. This research project used the Sri Lanka Institute
used to localize and the feature mapping. At the end of the of Information Technology at 7th floor as a location. The area
decision making process indoor map will be generated using consists rooms, corridors, and room doors and the furniture.
“IndoorOpenStreetMaps” (IndoorOSM). When examining the The following figure (Fig. 6) shows the screenshot of the
literature, authors have found multiple techniques to store the created map using JOSM which is available in the database, so
coordinates connected with the maps. Since IndoorOSM image of the downloaded indoor floor plan from the
provides a well-tested approach in mapping, it will be used accessBIM database. The fig.7 depicts the accessBIM
during the map generation step. database that including JOSM code to draw the maps.
.

148 2014 International Conference on Contemporary Computing and Informatics (IC3I)

Using MySQL commands, queries the results with regard to
an indoor environment. Fig 8 and 9 depicts the table that the
data consist in the tested location.

For testing purposes when the user selects the indoor

navigation option and selects the particular place where the
user wants to reach. The stored Geo-JSON data will be
retrieved by the application and convert this data into a string
type, then, that data will be passed to the HTML file and in
there with the help of Java Leaflet library [18], the relevant
map will be loaded into the application and when user zoom
the map users can identify the objects that are inside that
particular place. When the user is moving inside the building, Fig. 8. Indoor Places
distant and direction data will be saved in the database.

Fig. 9. Indoor Building

Fig. 10. Outdoor Navigation Results Fig. 11. Outdoor Navigation Results

Fig. 6. Downloaded Map from AccessBIM In the outdoor navigation situation for testing purposes
authors create the API in order to access the Google Maps [19].
Then retrieve the longitudes and latitude of the destination
buildings which are stored in the database. Then retrieve the
user’s current location and check for the last saved location
from GPS location service. Call the URL to get polyoverview
lines and extract in order to decode the encoded string. JSON
map gives JSON objects and pass it to an array to initialize the
map and it will update the marker position according to the
user's desires.
This proposed research the authors used outdoor maps to
identify and reach the locations. In the fig 10.dipicts the how
the device identify the location before entering an indoor
environment.
Fig .7. AccessBIM Database
The suggested method can overcome multiple user behavior.
As an example if the vision impaired individual entered to the
location using another door even it can be managed.

2014 International Conference on Contemporary Computing and Informatics (IC3I) 149

Fig.11. illustrates the results of outdoor navigation that depicts [5] Smith, P. A. Tracking Progress on Path for Indoor Navigation. Bachelor
how the individual reached the location and how to identify the thesis, RWTH Aachen, April 2011.
exact location of the building to provide correct location details [6] Fabritius, G. F. BurrowView - Path Reconstruction and Localization in a
Subterranean Animal Habitat Observation System. Diploma thesis,
and the map information to assist indoor environment. RWTH Aachen, September 2009.
[7] Bailey, T., and Durrant-Whyte, H. Simultaneous Localization and
Mapping (SLAM): Part II. Robotics & Automation Magazine, IEEE 13,
VI. CONCLUSION & FUTHER WORK 3 (2006), 108-117.
[8] Durrant-Whyte, H., and Bailey, T. Simultaneous Localization and
This paper presents an approach to create an indoor map that Mapping: part I. Robotics & Automation Magazine, IEEE 13, 2 (2006),
can be used to support the vision impaired people to navigate 99{110.
using their smart phones. Through the proposed model, it [9] Shin, H., Chon, Y., and Cha, H. SmartSLAM: Constructing an Indoor
demonstrations the ways of capturing data to generate maps Floor Plan using Smartphone. Tech. rep., Yonsei University, Tech. Rep.,
and the further enhancement of the algorithm towards error 2010. MOBED-TR-2010-2.
correction. The key aspect is to build an open, participatory and [10] Scholz, I., and OpenStreetMap community. Java OpenStreetMap Editor,
collaborative system through which users can contribute January 2012. http://josm.openstreetmap.de/, accessed 2014-05-23.
environment information using their mobile devices (crowd- [11] OpenStreetMap community. PicLayer Plugin for Java OpenStreetMap
Editor, Feburary 2012.
sourced). To address the issues involved in the application of http://wiki.openstreetmap.org/wiki/JOSM/Plugins/ PicLayer, accessed
collaborative indoor mapping, this paper discusses four 2014-05-12.
processing stages: data pre-processing, data smoothing and [12] Pomp, A. Indoor Navigation - Comparing Di_erent Indoor Location
error correction, data storage and “AccessBIM”. Determination Approaches. Seminar, July 2011.
[13] J.A.D.C.A. Jayakody, N. Abhayasinghe, I. Murray, “AccessBIM Model
The application introduced in this work is currently in for Environmental Characteristics for Vision Impaired Indoor
prototype phase, and there are many possible paths of future Navigation and Way Finding,” in International Conference on Indoor
work. The presence of more floor plans, wider range of Positioning and Indoor Navigation, November 2012, [Online].
supported mobile platforms, and a better indoor positioning Available:http://www.surveying.unsw.edu.au/ipin2012/proceedings/sub
system are obvious engineering improvements. Currently, the missions/
only type of indoor positioning that has any sort of [14] Pomp, A. IndoorNavigation - Comparing Di_erent Indoor Location
Determination Approaches. Seminar, July 2014.
infrastructure available at the campus area is positioned based
[15] Convert XML to JSON in PHP [EB/OL],
on wireless LAN signals. In future, someone have to focusses
http://www.ibm.com/developerworks/xml/library/x-xml2jsonphp/.
to convert the spatial information to voice that can be [16] Introducing JSON [EB/OL], http://www.json.org
understand by the vision impaired people. When vision [17] Getting Started with GEO,CouchDB & Node.js by Mick Thompson
impaired individual visit the places in the real world, the map (O'REILLY). Copyright 2011 David Thompson, 978-1-449-30752-3
can help them to navigate the existing places, and enhance their [18] An Open-Source JavaScript Library for Mobile-Friendly Interactive
experience of the indoor environment and the real world. Maps online avalable , http://leafletjs.com/ [accesed on 12.07.2014]
[19] Google Maps online avalable ,
An area of the work that requires great deal of development http://manuals.matrix.squizsuite.net/google-maps/chapters/google-map
is the process of producing indoor maps and route graphs out [accesed on 20.07.2014]
of the CAD format floor plans. The current process requires a [20] Nimsiri Abhayasinghe, Iain Murray ,A novel approach for indoor
lot of manual work, and is thus laborious to use and Indoor and localization using human gait analysis with gyroscopic data in
Outdoor Mobile Navigation susceptible to errors. Developing international conference on indoor positioning and indoor navigation,
november 2012, [online].
an algorithm to automate the process of extracting the map and Available:http://www.surveying.unsw.edu.au/ipin2012/proceedings/sub
the routing graph is thus an important goal. missions
[21] Nimali Rajakaruna, Iain Murray , Indoor positioning and indoor
ACKNOWLEDGMENT navigation efficient and adaptive generic object detection method for
This work has been supported by the Curtin University, Perth, Western indoor navigation in international conference on indoor positioning and
Australia and Sri Lanka Institute of Information Technology, Malabe, Sri indoor navigation, october 2013, [online].
Lanka. Available:https://www.researchgate.net/publication
[22] J. A. Bitsch Link, P. Smith, and K. Wehrle, “Footpath: Accurate
REFERENCES mapbased indoor navigation using smartphones,” in Indoor Positioning
and Indoor Navigation (IPIN), 2011 International Conference on, 2011.
[1] IndoorOSM. Available online:
http://wiki.openstreetmap.org/wiki/IndoorOSM (accessed on 10 April
2013).
[2]. Goetz, M.; Zipf, A. Extending OpenStreetMap to Indoor Environments:
Bringing Volunteered Geographic Information to the Next Level. In
Proceedings of the Urban and Regional Data Management:Udms Annual
2011, Delft, The Netherlands, 28–30 September 2011;pp. 47–58.
[3] K. Kaemarungsi and P. Krishnamurthy, "Modeling of Indoor Positioning
Systems Based on Location Fingerprinting," in Proc. Of IEEE InfoCom
2004, Hong Kong, China, March, 2004.
[4] Bitsch Link, J. A., Smith, P., and Wehrle, K. FootPath: Accurate Map-
based Indoor Navigation Using Smartphones. In International
Conference on Indoor Positioning and Indoor Navigation (IPIN), 2011
(2011).

150 2014 International Conference on Contemporary Computing and Informatics (IC3I)