A Pilot Study: Shuttle Bus Tracker App for Campus

Users
Su Mon Chit, Lee Yen Chaw, Chee Ling Thong and Chiw Yi Lee
UCSI University, Kuala Lumpur, Malaysia
chitsm@ucsiuniversity.edu.my
chawly@ucsiuniversity.edu.my
chloethong@ucsiuniversity.edu.my
leecy@ucsiuniversity.edu.my

Abstract—Many institutions of higher learning provide able to reach directly to their preferred destination such as
shuttle bus service to students and staff to and from between school or universities and often send them to nearest
public pick up point and campus for their convenience. interchange point. This controversy issues often disappoint the
However, such service is not receiving positive feedback as users students to use public transport. Hence, some of the institutes
have to wait without any clue when the bus is going to arrive. or universities provide shuttle bus services between public
Furthermore, the time schedule posted on the University website interchange point and respective universities or institutes.
is often inconsistent. Hence, providing the accurate bus arrival Although, shuttle bus services solve traffic issues for the
time is important as it can reduce waiting time at bus stops as students, often it disappoints students because of arrival time
well as to decide on the travel plan. The proposed shuttle bus
and waiting time to take shuttle bus. University often receive
tracker app aims to minimize waiting time by incorporating
features that allows commuters to know the location of the bus in
negative feedback from students due to excessive waiting time
real-time, to make query about the arrival time of the bus as well and unreliable bus schedule [12,13]. According to Yusoff,
as to provide real-time traffic update about the shuttle bus McLeay and Woodruffe-Burton (2015), student support
routes. The app calculates the bus arrival time by getting the facilities have been identified as one out of the 12 factors that
location of the bus chosen and the bus stop then passing their drive student satisfaction in higher education [2]. Since,
coordinates to Google's distance matrix Application providing shuttle bus service can be considered one of the
Programming Interface (API) web service which returns the time student support facilities, thus it is essential for administrator of
to travel from the selected shuttle bus location to the chosen bus the University to take this matter seriously for strategic
stop. The Google Maps Distance Matrix (API) is a service that planning.
provides travel distance and time for a matrix of origins and
destinations, based on the recommended route between start and As mentioned earlier, the accuracy of bus arrival time is
end points. In order to achieve the main aim of this study, a pilot important information for commuters and failure to provide
test was conducted and test results suggest that the proposed app accurate bus arrival time or long waiting time at the bus stops
has achieved its objectives. often makes commuters reluctant to use bus service. Most of
the bus operating companies provide bus schedule at the bus
Keywords—mobile application; real-time location; smart stops or on their web for the commuters. However, only limited
phones; vehicle tracking system information such as operating hours and time intervals are
provided in their schedule. Hence, information such as real-
I. INTRODUCTION time schedule and arrival time of the buses may not be
available in their schedule. Buses arrival time calculation
Traffic plays an important role not only in developing depends highly on speed of the vehicle as well as traffic
country but also in developed country. People use it for many condition, which are also link to weather conditions and traffic
purposes such as travelling to work, school and personal use, incidents [1]. Therefore, it is more complex to predict travel
etc. Among them, the bus service is not only affordable for time and to get accurate real-time bus information without the
public transport passengers (also known as commuters) but help of technology. With the help of advanced mobile
also reduce the fuel consumption and traffic congestion caused technology, smartphones can be used to track the real time
by private car users. However, limitation of platforms which location of the buses and estimate bus arrival time.
provide traffic information leads the increments in traffic
congestion. The bus arrival time is important information to Nowadays, phones can be used not only to make call or
most commuters and longer waiting time for the bus often send short message, but also for smart services such as
discourage the commuters and make them reluctant to take browsing internet, taking photos or videos, working on
buses. In order to overcome the traffic congestion caused by spreadsheet and many more. Hence, these are called
limited traffic information, most of the governments provide smartphones. Smartphones are one of the popular medium to
support not only to public traffic systems but also the traffic track the real-time information such as real-time bus
information systems [1].Students are one of the main crowds, information as those are portable to the users and easy to use
among all the commuters, who are heavily depending on public [10]. Besides, smartphone applications or mobile applications
transport service. However, the public transport might not be (also known as apps) are growing trend for modern world

the copyright clearance code notice is:

978-1-5090-6255-3/17/$31.00 ©2017 IEEE
today [8]. Grab or MyTeksi is one of a good example app Real-time web-based vehicle tracking using GPS consists
which provides services for car and taxi booking. Commuters of two major units; built-in unit and sever/central monitoring
can use this app to engage car or taxi service and track the real- station [6]. In-vehicle unit is installed into the vehicle and
time location of car or taxi [11]. The app is an application or a capture information such as real time location of vehicle,
piece of typically a small, specialized program or software speed, status of door open or close, ignition on/off status. It is
designed for a particular purpose downloaded onto mobile also responsible for transmitting information to tracking server
devices. Therefore, it helps users in a way of reducing waiting which can be located anywhere in the world. In-vehicle unit
time and effort to use public transport services. uses GPS receiver to capture current location and vehicle
speed, Central Processing Unit (CPU) to process the raw data
As mentioned before, smartphones can be used as tracking collected from GPS to provide location and speed information.
system for bus location as well as to estimate bus arrival time It is also responsible for checking door open/close status and
for commuters. The pilot test in this study focuses on pilot ignition on/off status. Due to the nature of web-based system,
testing the mobile app among the commuters who are engaging the system will have slower response time for the users
bus service in the campus. In order to achieve the main compare mobile application.
objective of the study, the pilot test was conducted in a private
university located in Kuala Lumpur, Malaysia. There are key Another web based GPS-GPRS(General Packet Radio
functionalities of the mobile app as it allows shuttle bus users Service) vehicle tracking system was designed and
to track the real-time location of a shuttle bus as well as real- implemented for enterprises owner to collect present and past
time traffic status. Besides, studies are also carried out to location information of the target vehicle on Google Map
identify the best techniques to fulfill the user requirements and through purpose designed website [6]. Location information of
functionalities of the system and to be presented as a result of the vehicle was collected from GPS device which is installed
pilot test of the project. in the target vehicle and send using GPRS service provided by
GSM (Global System for Mobile communication) network.
II. EXISTING VEHICLE TRACKING SYSTEMS Hence it uses GPS receiver and GSM modem and installed in
the vehicle. The movement of the vehicle can be tracked via
Google Map by using web based application.
Public transport providers usually make vehicle timetables
available online for commuters [3]. However, limited B. Mobile-based Vehicle Tracking Systems
information such as operating hours, time intervals are
provided in the timetables. In order to lessen the problems Mobile phone based roadway transport system is developed
faced by commuters, a number of researches have been using bus passengers participatory sensing [5]. It collects users’
conducted to develop various vehicles tracking systems around information to estimate the bus travelling routes and predict the
the world [4]. The following sections discusses on some of the arrival time at various bus stops along the route. Participation
vehicle tracking systems which have been developed all over from users is the main information which has been used in this
the world. system and smart phones can be used to collect users’
information. Hence, high risk will be imposed to users’
A. Web-based Vehicle Tracking Systems personal information while collecting the data for the system.
Combination of neural network and Kalman filter
techniques is one of the effective methods to predict expected
EasyTracker is introductory application for transit tracking
bus arrival time at different bus stops along a service route [1].
The technique uses historical data and current Global and arrival time prediction in smaller transit agencies [7]. It is
Positioning System (GPS) measurements. This means, it required to obtain the smart phone in order to use the
considers historical data together with real time location of the EasyTracker. Application should then be installed in
bus to calculate expected bus arrival time. Although, the smartphones and place a phone in each vehicle. Data collected
system provides satisfactory performance and accurate from GPS traces has been used to automatically determine
prediction of bus arrival times, it requires additional routes served, locate stops and infer schedules.
information such as historical data which may not be available
in certain situation. Another approach which uses machine learning technique
to predict departure time of a mobile user via mobile phone
Another bus arrival time prediction systems has been sensor [10] has been developed. It uses the machine learning
developed using passengers participatory sensing [3]. approach for predicting the departure time of a mobile user.
Passengers’ information such as surrounding environmental The approach is based on three information which mobile
context is effective to utilize in estimating bus travelling routes phone produces; location, accelerometer and time. However,
and predicting bus arrival time at various bus stops. The system it cannot trace the location of the vehicle as well as traffic
consists of three main components namely (i) sensors to share status and arrival time.
information to users, (ii) querying information from users and
(iii) server to collect information from users and processing
C. Bus Arrival Time Estimating Systems
collected information [3]. It is mentioned that commuters are
willing to share their location information in order to calculate The accuracy of Bus traveler information systems (BTIS)
the bus arrival time; hence it breaches the commuters’ personal depends on several factors and existing BTIS in India does not
information. consider the real time data and quality control of the data which
can improve the performance of the underlying prediction performs the best among the four proposed models for
system of bus arrival time [14]. There are several studies which predicting the bus arrival times at bus top with multiple routes.
predict real time arrival time of the bus however it does not However, their current research only focus on current traffic
fulfill the requirement of India traffic system as the traffic in conditions of bus data and other factors such as running times
India is much different. Most of the advanced transport systems of vehicles or traffic flow variation are not counted in the
in India are rudimentary and concentrated more on technology research. Hence, it may not able to provide accurate real time
demonstration. One of the main factors which make the delay bus arrival information to the users.
of bus in India is due to the waiting time or delay time at bus
stops. This delay makes significant effect to the accuracy of Providing accurate bus arrival time to passengers’ mobile
bus travel time however most of the current studies have not devices usually helps them in planning their travel time as well
look into this matter yet. Hence another study has done by as save their dwelling time at bus stops. It can not only helps
using model-based algorithm which utilize real time data of the passengers but also help the operators to monitor their
users and takes delays automatically into account for an operation schedule, react instantly and evaluate the operation
accurate prediction of bus arrival time. A very well-known efficiently. Another research has been proposed to predict bus
model-based approach Kalman filtering technique has been arrival time by using historical data and real-time situation
information. There are two stages in their analysis. Firstly,
used in this study and the approach divided the total travel time
of a vehicle into two components – ‘running time’ and ‘delay Radial Basis Function Neural Networks (RBFNN) model
time’ and analyzed them separately. In this study, a GPS unit which is used in order to learn and approximate the nonlinear
was inbuilt with a GPRS modem which sent the location relationship in historical data in the first phase and secondly
details to the remote server to actual calculation. Hence, it online oriented method in order to adjust the actual situation
required an extra device to calculate bus arrival time. [18]. The idea of this is to supplement or modify the result
predicted using RBFNN model with the practical information.
Another research has been conducted to estimate bus arrival Several other models such as multiple linear regression models,
time by using historical data which is based on Automated BP Neural Networks and RBFNN without online adjustment
Passenger Counter (APC) data [15]. Prediction model for travel are also used to compare the result and to get the most accurate
between two adjacent bus stops and an algorithm to search for result. Based on their result, the approach which uses RBFNN
a next arriving bus are involved in this study. The drawbacks of together with online adjustment has a better performance in
predicting travel time on historical data based models is the prediction bus arrival time. However, the reliability of the
traffic patterns are cyclical and the ratio of the historical data result given by this approach for predicting bus arrival time
versus real time travelling time will remain constant. The will need to be further studied in order to provide solid and
conventional approaches such as regression models limit the accurate result.
nature of traffic as constant and hence it makes the prediction
inaccurate in most of the cases. As bus-arrival-time information service has been one of the
important components of the advanced traveler information
Complex traffic conditions are one of the reasons why it is system (ATIS), a new bus travel time prediction model has
difficult to estimate bus arrival time and often lead to been proposed. Bus arrival time information service is
inaccurate calculation of bus arrival time. However, accuracy important to attract additional ridership, higher customer
of bus arrival time is crucial for passengers in order to make satisfaction and positive psychological factors. In this paper, a
travel plan or decide their choice for transits. A system which short-term prediction model is proposed by using inputs of
calculates bus arrival time based on smart card data has been real-time bus location data collected by the Automatic Vehicle
developed as most of the bus arrival time are usually estimated Location (AVL) devices and the RFID (Radio-frequency
using the boarding time of the first passenger at each station. identification) data simulated by manual collected data [19].
However, these are not accurate in a way as many passengers RFID data based models have the advantages of generalization
usually swipe the card much before their alighting. Hence, in under various conditions and will have a wide range of
their study, a combination data collected from swiped smart applications in bus travel time prediction. However, it required
card with the actual bus arrival time by the manual survey data special RFID devices as well as need to use manual collected
[16]. The approach used in this research is good as it involved data hence it will not be effective to provide bus arrival time
manual survey data however it may not be efficient in a long information service in long term.
term. Besides, the data which has been collected from swiped
smart card will not be accurate as expected because of user Table 1 shows the comparison of different types of vehicle
behavior. tracking system discussed earlier.

Another search has been conducted to provide accurate bus TABLE I. COMPARISONS BETWEEN DIFFERENT VEHICLE TRACKING
arrival time information to the passengers in order not only to SYSTEMS
reduce their anxieties and waiting time at the bus stop but also
to improve management and service level [17]. There are Name of the Authors Retrieve
System Required Mobile d user
various approaches to calculate the most accurate bus arrival Special Phone location
time and the researchers use several method such as vector Device based informati
machine (SVM) artificial neural network (ANN) and many on
more. The research has been conducted in transit-oriented city, Online Bus M. Zaki, I.
Hong Kong where they use many different routes for the same Arrival Time Ashour, Yes No No
Prediction M.Zorkany,
bus stop in urban areas. They have concluded that SVM model
Name of the Authors Retrieve Name of the Authors Retrieve
System Required Mobile d user System Required Mobile d user
Special Phone location Special Phone location
Device based informati Device based informati
on on
Using Hybrid and B. travel time Teng Jing,
Neural Hesham basing on both Chen Guojun,
Network and GPS and and Shu
Kalman filter RFID data Qichong
Techniques
Predicting Bus Pengfei Zhou,
Arrival Yuanqing III. METHODOLOGY
Timewith Zheng, and
Mobile Phone Mo Li
Yes Yes Yes
based
Participatory
Sensing

Real-Time
Bus Tracking Shruti
Android Kotadia, No Yes Yes
Application Ankita Mane,
Jignasha
Dalal
Mobile Phone K.Sedhurama
based n, and Mrs.
Roadway R.Kavitha
Transport
No Yes Yes Fig. 1. High Level Architecture of Shuttle Bus Tracking System
System using
Participatory
Sensor
The high level architecture of shuttle bus tracking
Development R.P.S. system has been shown in Fig. 1. Android smart phone
of a real-time Padmanaban, (Version 5.0 and above is used to installed the app for
bus arrival K. Divakar1
prediction L.
driver as well as commuter. Mobile phones can be used
Yes Yes No
system for Vanajakshi, not only to make call or send short message, but also for
Indian traffic and S.C. smart services such as browsing internet, taking photos or
conditions Subramanian
videos, working on spreadsheet and many more. The
Bus Arrival Shaowu mobile phone used by shuttle bus driver will be mounted
Time Cheng, Baoyi on the shuttle bus and the application is installed in it.
Prediction Liu, and
Model Based Botao Zhai
Yes No Yes Simple user interface for shuttle bus driver is provided to
on APC Data log in to the system as well as to update real-time
Bus arrival Yuyang
location and traffic information. The current location of
time Zhou, Lin shuttle bus will be automatically obtained via Google's
calculation Yao, Yanyan API and in-vehicle smartphone is automatically
model based Chen, Yi Yes No Yes
on smart card Gong, and transmitting its GPS coordinates to a central location
data Jianhui Lai server, using a cellular data link. Data collected from GPS
traces has been used to automatically determine routes served,
Bus arrival Bin Yu,
time William H.K.
locate stops and infer schedules [7]. As shown in Fig. 1,
prediction at Lam, and Mei shuttle bus user or commuters will also need to install
bus stop with Lam Tam Yes No Yes the application in their mobile in order to track the
multiple
routes current traffic condition as well as the bus location. The
updated location of the shuttle bus can be seen by
Bus Arrival Lei Wang, commuters once the commuter login to the application.
Time Zhongyi Zuo,
Prediction and Junhao The real time updated location of the bus and it’s last
Using RBF Fu
Yes No Yes seen will be shown on Google Map as well as the bus
Neural
Networks
driver name. Users who log in as commuters also need
Adjusted by to turn on the mobile data to see the latest location of the
Online Data shuttle bus as well as last seen on the shuttle bus and its
Predicting bus Song
real-time Xinghao,
Yes Yes Yes driver’s name. Besides, the commuter will be able to
check the estimated duration to wait for shuttle bus A. Profile of Respondents
arrival. Moreover, smartphones can be used to collect
users’ information such as users’ current location [5]. TABLE II. DEMOGRAPHIC STATISTICS

The application calculate the shuttle bus arrival time by

getting the location of the bus chosen and the bus stop Variable n % Cumulative
then passing their coordinates to Google's distance Gender
matrix API web service which returns the time to travel Female 3 18.8 81.3
Male 13 81.2 100.0
from the selected shuttle bus location to the chosen bus
stop. Status
Student 15 93.8 93.8
Staff 1 6.3 100.0
IV. PILOT TESTING ON PROTOTYPE
A prototype of an android mobile app for shuttle bus on Table II shows the basic demographics of a total of the 16
campus is developed based on the user requirements gathered shuttle bus mobile app users. In the sample, male represented
at the preliminary study conducted earlier. 81.2% while female 18.8%. The percentage of students
(93.8%) was very much higher than staff (6.3%). This is a
Simple and friendly user interface design is used in this common phenomenon where students are less affordable to
prototype. The stages for pilot testing for both shuttle bus have their own vehicle. Thus, depend heavily on public
driver and commuter are as follow: transport.

A. Shuttle Bus Driver

i. Driver needs to turn on the mobile data. B. Data Analysis
ii. Download and install Shuttle Bus Tracker App on
Android (Version TABLE III. MEAN SCORE - THE FEATURE OF A SHUTTLE BUS TRACKER
APP
5.0 and above) smart phone.
iii. Register as a bus driver (Using Staff ID, Username, No. Item Mean
Password, Contact Number and Email Address). The shuttle bus tracker app is able to show the
iv. Log in using “Username” and “Password”. 1. real-time location of a shuttle bus. 3.94
v. Real-Time location will be automatically updated.
vi. Shuttle bus driver will able to update real time traffic The shuttle bustracker app allows me to make
2. 3.69
condition: “Traffic” or “No Traffic”. query about the arrival time of a shuttle bus.

The shuttle bus tracker app is able to provide

B. Shuttle Bus Commuter real-time traffic updates on a shuttle bus
i. Commuter needs to turn on the mobile data. 3. 3.69
routes.
ii. Download and install Shuttle Bus Tracker App on
Android (Version 5.0 and above) smart phone.
iii. Register as a user (Using Student ID, Username, All the 3 items shown in Table III are measured using 5 point
Password, Contact Number and Email Address). Likert Scale where 1= Strongly Disagree, 2 = Disagree, 3 =
iv. Log in using “Username” and “Password”. Neither Agree nor Disagree, 4 = Agree and 5 = Strongly
v. Real-Time location of the bus and driver name will be Agree.
seen on Google Map.
vi. User will able to check the estimated waiting time of Item 1 for Table III has a mean score of 3.94 whereas items 2
the shuttle bus from current location of the bus to and 3 both have the same mean score of 3.69.
current location of the user.
VI. DISCUSSION
V. SURVEY Based on the mean score obtained for each of the features
A total of 16 shuttle bus users were provided with a link to of the said app, the results indicating that the shuttle bus users
download the shuttle bus tracker app. The sample size of the are gnerally agreeable that the app can show the real time
pilot test is considered sufficient as stated by Fink (1995b) that location of a shuttle bus, the app allows them to make query
the minimum number for a pilot test is 10 [20]. Once the about the arrival time of a shuttle bus and is able to provide
shuttle bus users had tried the stated app, they were asked to real-time traffic updates on a shuttle bus routes. During the
provide their feedback by rating the features of the app based pilot testing phrase, inconsistencies were found due to
on the 5 point Likert scale. The survey was carried out via an database issues which sometimes make the drivers/commuters
online mode. not able to log in to the system. The development team has
fixed and refined the interface of the system.Since this is a
pilot testing with a small number of sample size, the next
course of action is to conducting an actual testing for a larger
group of shuttle bus users on campus in order to draw a [6] Dr. Khalifa A. Salim, and Ibrahim Mohammed Idrees, “Design and
implementation of web-based GPS-GPRS vehicle tracking system”. In
conclusion on the usefulness and the general acceptance of the
IJCSET, 3(12), 2013.
shuttle bus tracker app before it is implemented throughout [7] James Biagioni, Tomas Gerlich, Timothy Merrifield, and Jakob
the campus. Eriksson, “EasyTracker: Automatic transit tracking, mapping, and
. arrival time prediction using smartphones”. SenSys’11, November 1–4,
2011.
VII. CONCLUSION [8] Laudon, K. C., and Laudon, J. P. (2016). Management Information
Systems: Managing the Digital Firm (14th ed.). Essex, England:
In this paper, shuttle bus tracker app, an android mobile Pearson.
app prototype, for campus users has been developed to track [9] Biagioni, J., Gerlich, T., Merrifield, T., and Eriksson, J.: “EasyTracker:
automatic transit tracking, mapping and arrival time prediction using
real-time location of the bus , to make query about the arrival smartphones. In Proceedings of ACM SenSys, pp. 1-14, 2011.
time of the bus as well as to provide real-time traffic update [10] Ron Biton, Gilad Katz, and Asaf Shabtai,”Sensor-based approach for
about the shuttle bus routes. In order to achieve the main predicting departure time of smartphone users”. 2nd ACM International
objective of this study, a pilot test was conducted in this study Conference on Mobile Software Engineering and Systems, 2015
and the test results indicates that the proposed app is capable [11] Grab, https://www.grab.com/my/
of providing real time traffic update and location of the shuttle [12] Bus Comments and Complaints,
http://www.surrey.ac.uk/currentstudents/campus/transport/bus/commen
bus thus predicting bus arrival time for shuttle bus users and ts/
allow them to make query about the arrival time. [13] Bus Service Feedback,
https://www.warwicksu.com/campaigning/campaigns/transport/buses/r
ACKNOWLEDGMENT eporting/
[14] R.P.S. Padmanaban, K. Divakar1 L. Vanajakshi, and S.C.
This research work is supported by UCSI University under Subramanian, “Development of a real-time bus arrival prediction
the University Pioneer Scientist Incentive Fund (PSIF). system for Indian traffic conditions”, IET Intell. Transp. Syst., 4(3), pp.
189–200, 2010.
REFERENCES [15] Shaowu Cheng, Baoyi Liu, and Botao Zhai, “Bus arrival time
prediction model based on APC data”. The Sixth Advanced Forum on
Transportation of China, 2010.
[1] M. Zaki, I. Ashour, M.Zorkany, and B. Hesham, “Online bus arrival
[16] Yuyang Zhou, Lin Yao, Yanyan Chen, Yi Gong, and Jianhui Lai, “Bus
time prediction using hybrid neural network and Kalman filter
arrival time calculation model based on smart card data”.
techniques”, International Journal of Modern Engineering Research
Transportation Research Board 95th Annual Meeting, 2016.
(IJMER) vol.3, pp. 2035—2041, 2013.
[17] Bin Yu, William H.K. Lam, and Mei Lam Tam, “Bus arrival time
[2] Yusoff, M., McLeay, F., and Woodruffe-Burton, H., “Dimensions
prediction at bus stop with multiple routes”. ELSEVIER 19(6), pp.
driving business student satisfaction in higher education, quality
1157–1170, 2011.
assurance in education”, 23(1), pp. 86-104, 2015.
[18] Lei Wang, Zhongyi Zuo, and Junhao Fu, “Bus arrival time prediction
[3] Pengfei Zhou, Yuanqing Zheng, and Mo Li: “How long to wait?: using RBF neural networks Adjusted by online data”. ELSEVIER vol.
Predicting bus arrival time with mobile phone based participatory 138, pp. 67-75, 2014.
sensing”. In MobiSys’12, June 25–29, 2012. [19] Song Xinghao, Teng Jing, Chen Guojun, and Shu Qichong, “Predicting
[4] Muruganandham, and P.R.Mukesh, “Real time web based Vehicle bus real-time travel time basing on both GPS and RFID data”,
Tracking using GPS”, World Academy of Science, Engineering and ELSEVIER vol. 96, pp. 2287-2299, 2013.
Technology, 2010. [20] Fink, A. “How to ask survey questions”, Thousand, CA:Sage, 1995b.
[5] K.Sedhuraman, and Mrs. R.Kavitha, “Mobile phone based roadway
transport system using participatory sensor”. In International Journal
On Engineering Technology and Sciences – IJETS 2(4), 2015.