A novel method for Smart school bus tracking system using Machine learning and IoT

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 07 | July 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 2233
A novel method for Smart school bus tracking system using Machine
learning and IoT
Aiswarya Kannan1, Luckneshwaran.E 2, Ranjith.P3, VijayaKumar.S4
1Assistant Professor, Department of ECE, SRM TRP Engineering College, Tiruchirappalli, India
2,3,4 Third year SRM TRP Engineering College, Department of ECE, SRM TRP Engineering College, Tiruchirappalli,
India
----------------------------------------------------------------------***---------------------------------------------------------------------
Abstract
Thousands of students use school transportation around
the world. We require a transportation system that is
effective, reliable, safe, and smart. The suggested system
depicts a transportation concept that provides real-time
tracking, calculates optimal routes to destinations,
detects intrusion, and assists in the maintenance of
transportation system statistical data. IR sensors and
RFID tags are used to create an IoT network. The
detection of intrusion is done via facial recognition.
Google Maps, GPS, and accelerometer data are used to
detect live location. A bus-mounted Raspberry Pi
microcontroller interfaces with a centralised Firebase
cloud platform. Admins and parents can access the
mapped data via a mobile application. The system saves
important data such as driving abilities, attendance
analysis, and the optimal routes. In the ML optimizer, the
data is effectively used.
Key Words: IoT, IR sensor, RFID, Raspberry Pi, GPS,
Cloud, Google Maps, ML
1. INTRODUCTION
For parents, getting their children to and from school is a
major problem. Several concerns arise, including
extended waits for delayed buses, kidnapping, and kids
deboarding at incorrect stations. Other concerns include
a lack of knowledge regarding the routes used by the
school bus to reach its destinations, as well as stopping
unauthorised individuals from accessing the bus. Student
safety has long been a top priority, especially for
working parents, whose solutions must be continually
enhanced [1]. Drivers may not be able to identify all
students on time if one goes missing. Children are
frequently prohibited from using cell phones on school
grounds. It is difficult for school officials to contact each
parent about their child's safety. [2]. If the motorist is
involved in rash driving, the driver should be assisted
and the caretakers should be informed [3]. By tracking
and monitoring students, the Internet of Things provides
a much better approach for ensuring their safety.
IoT is implanted as the primary concept to solve
and overcome these challenges. Web-based bus tracking
is provided by the proposed solution. Students entering
the bus are detected by an IR sensor at the entry, which
is verified by an RFID reader. The RFID device is scanned
for identification and information retrieval. When
students board the bus, they must wear their unique tag
to be scanned by the RFID scanner. Parents and
administrators are notified when their child boards the
bus using confirmed scanning. The Raspberry Pi
microcontroller is used to update attendance over
Firebase. If a scan mismatch occurs, a Firebase notice is
issued to detect an intruder and capture the intruder's
photo. When boarding or disembarking from the bus,
each child is identified.
The accelerometer keeps track of the bus's
condition and analyses how it drives. In real-time, the
technology monitors the child's location and bus route.
The car is followed by the GPS module.The GPS position,
driver behaviour, and emergency conditions are used to
monitor the overall system. Caretakers would be
contacted right away if an accident or emergency
occurred. All preceding sections would be saved in
Firebase for future use.
1.1 Literature Survey
F Judy et al. [4] proposed a school bus
monitoring system that uses RFID and GPS to
communicate to a remote server using Wi-Fi and an
ESP8266 microcontroller. Caretakers could access
information utilising the cloud-based Firebase
messaging service. Jisha et al. [5] developed a car
monitoring system for schoolchildren that employed GPS
and GPRS/GSM technology to ensure the safety of the
students. The system comprised of an Internet-enabled
android application that communicated with a server.
W. Pattanusorn et al. [6] devised a system that
automatically registered children's information at the
entry-level when they passed through the scanner. SMS
alerted parents of their children's school bus arrival and
departure times. H Eren et al. [7] proposed a sensory-
based driving behaviour analysis technique. The tracking

was done using a smartphone to simulate a car-
independent system without the need for vehicle-
mounted sensors. Muneer et al. proposed an android-
based framework that combined GPS data with Google
maps data to accurately locate the misplaced mobile
phone. [8]
The need for the development of a real-time
transportation-based information system for users,
which might aid in improved trip planning and reduce
bus waiting time, sparked interest. Real-time data
processing can help commuters get at their destination
faster by reducing waiting time.
The proposed work uses IoT with RFID,
Raspberry Pi, IR sensor, GPS, Firebase, and Google Maps
to achieve features such as attendance analysis, notifying
end-users with alighting and boarding alerts, location
tracking through GPS by integrating Google Maps,
intrusion and accident detection, and cloud storage to
expand the security framework and traceability aspects
of the child. The mobile application would make this
information available to the administrator, parents, and
driver.
2. COMPONENTS
The proposed system utilizes the following
hardware and software components:
2.1 Hardware Specifications
The bus's hardware is the most crucial
component. The hardware utilised for prototype
development is described in this section, which
comprises the following:
A. IR Sensor
The primary function of an IR Sensor is to measure and
detect infrared radiation in the surrounding
environment. The proposed system uses an IR sensor as
the first step in verifying the student's entry into the bus.
With a pair of infrared transmitters and a receiver tube,
an IR Sensor module has an adaptable potentiality of
atmospheric light. Infrared technology is used in a
variety of wireless applications.
B. RFID Reader RC522
The RFID RC522 Card Reader Module, based on the
MFRC522 controller, is a low-cost 13.56 MHz RFID
reader module. The module necessitates a 3.3V power
supply. It can communicate with any CPU board directly
using the SPI protocol, and it also supports I2C and
UART. It is utilised for attendance analysis and person
identification in the proposed system.
C. Raspberry Pi 3 B+ Micro-controller
The Raspberry Pi foundation microcomputer that was
created to promote programming and computing
principles. It has a 64-bit quad-core processor with a
clock speed of 1.4GHz and dual-band 2.4GHz. It has 5GHz
wireless LAN and Bluetooth connectivity, making it an
ideal alternative for highly networked designs. Its high
processing power and on-board connectivity make it
ideal for IoT applications.
D. Camera Module
The 5-megapixel Camera Module Rev 1.3 is a specially
developed Raspberry Pi add-on. A unique CSI interface is
utilised for camera interaction. The CSI bus provides
extremely high data speeds and consistently transports
pixel data. It is used to capture a snapshot of the intruder
in the proposed system.
E. GPS Module –Neo 6M
The Neo-6M GPS module is a reliable GPS receiver with a
25 x 25 x 4mm ceramic antenna built in. It has a good
satellite search capacity. The power and signal indicators
can be used to check on the module's status. It is used to
gather information about geographical parameters.
F. MPU6050 Accelerometer
The MPU6050 is a single-chip 3-axis accelerometer and
gyroscope. It is also known as a six-axis motion tracking
or six Degrees of Freedom (DoF) device because of the
three accelerometer and three gyroscope outputs.
Hardware components used in the prototype
development are shown in Figure 1.
Fig -1: HardwareComponents
2.2 Specifications for Software
HTML, JS, CSS, jQuery, and Bootstrap were used to create
the android application. In order to view the current
location, the Google Maps API has been integrated into
the programme. Using Firebase fire-store, we were able
to get real-time changes. Apache Cordova was used to
encapsulate this view in an app.

A. Firebase
The Firebase database is a cloud-based database that
allows for real-time data syncing and storage across
users around the world. It promotes user participation
and serverless application development. JSON is the
format in which the NoSQL database is stored. When
developing cross-platform programmes, a Realtime
Database instance can be shared among all clients. Even
if the app goes down, the data is still accessible.
B. API for Google Maps
Google has created application programming interfaces
(APIs) that allow users to communicate with Google
Services and integrate them with other services.
Analytics, machine learning as a service, and user data
access are among the features. Google Maps can also be
integrated into a website or application. It is possible to
provide users with relevant content and to customise
their map view according to the site. It is used in the
suggested model to add a map to the Android app to
detect bus routes and deliver the best way to destination
with real-time traffic updates.
C. Node-RED
Hardware components, APIs, and other web services are
used to collaborate. Node-RED is a popular visual
programming tool that comes with a web-based editor.
JSON can be used to save the flows and share them with
others. They can be used at any moment with a simple
click. It is an event-driven and non-blocking approach
that is based on Node.js. The Raspberry Pi is being used
in the current system.
D. jQuery and Bootstrap
Bootstrap is a web application framework. Consistency is
maintained across browsers and device screen sizes
thanks to HTML and CSS. It has a large number of plugins
and themes. Because jQuery is an open-source tool, it is
used to create client-side scripting on HTML, which
makes websites more responsive.
E. Apache Cordova
Apache Cordova is a mobile development framework
that is free and open-source. By combining HTML5, CSS3,
and JavaScript, we can create hybrid web applications.
It's in charge of bridging the gap between web-based
apps and native mobile capabilities. It serves as a link
between web apps and mobile devices.
3. PROPOSED SYSTEM ARCHITECTURE
The suggested system seeks to provide effective services
through the integration of various technologies and the
Internet of Things. The suggested system's diagrammatic
representation is shown in Figure 2.
Fig -2: Representation of the System Proposed
In Figure 3, you can see a hardware prototype of the
system. It emphasises the various components that were
used to construct the prototype. The Raspberry Pi serves
as the system's brain, connecting to several sensors like
as GPS, Accelerometer, RFID, IR, and Camera. For
database management, it's also linked to Google
Firebase. The Raspberry Pi and the Firebase have most
of the data in sync.
The administrative, driver, and parent mobile apps are
designed to keep everyone up to date on the latest
information, such as routes, live monitoring, attendance,
and notifications.
Fig -3: Prototype Hardware
The accident detection method is based on the Random
Forest Machine Learning Classifier (Fig. 4). Random
Forest is a popular machine learning algorithm that
belongs to the supervised learning technique. This
algorithm solves classification and regression problems
in machine learning. It is based on the notion of
ensemble learning, which entails merging numerous

classifiers to solve a complex problem and improve the
model's performance.
Random Forest Algorithm in Action (Fig. 4)
"Random Forest is a classifier that contains numerous
decision trees on various subsets of a given data set and
takes the average to increase the predicted accuracy of
that data set," as the name suggests. Rather than relying
on a single decision tree with a majority of votes, the
random forest classifier collects predictions from all
trees to forecast the final result. The greater the number
of trees in the forest, the higher the accuracy and the less
likely it is to overfit.
The Random Forest algorithm, as shown in Figure 4,
works in two phases: first, it creates a random forest by
mixing N decision trees, and then it makes predictions
for each tree formed in the first phase.
The steps and diagram below show how the working
process works:
Step 1: Picking K data points at random from the training
set.
Step 2: Create decision trees for the data points you've
chosen (Subsets).
Step 3: Determine the number of decision trees to be
constructed (N).
Steps 4: steps 1 & 2 are repeated.
Step 5: Find the forecasts of each decision tree for new
data points, and assign the new data points to the
category with the most votes.
In Section 4, Fig-8 depicts the implemented architecture
algorithm that outlines the entire flow. It emphasizes
how the various elements (hardware and software)
interact with one another. Students enter the bus
through an IR sensor, which is verified by an RFID
reader.
Following successful verification, the tag data is
transferred to the cloud via a Raspberry Pi
microcontroller, where it can be accessed by admins,
parents, and drivers via a mobile application. This aids
parents and school officials in collecting attendance
reports. If an RFID reader is not discovered, the camera
is used to identify a suspected breach. At that point, a
photograph of the person is taken.
Throughout the journey, data from the accelerometer
and GPS module is collected and sent to the cloud.
The physical coordinates of the vehicle's location as well
as the speed can be tracked using GPS-based information
combined with Google maps data. Machine Learning
information from the driving behaviour analysis is used
to detect accidents, assuring student safety. In the event
of an emergency, the administration can contact the
driver.
4. RESULTS
This phase of the pilot study develops the
findings and conclusions. The bus route has three stops
in the design execution. The data was retrieved for two
months (February and March 2020). The traffic pattern,
driving behaviour, and identification of possible
bottlenecks in that route are among the conclusions
drawn from the study and testing. Data for the proposed
system was gathered using GPS units installed in school
buses.
The 3.6-kilometer experimental route was
chosen from Vidyalankar Institute of Technology in
Mumbai to Matunga Railway Station in Mumbai. On this
route, there are nine bus stops. GPS data from the
devices installed in the buses was relayed to a server
every 60 seconds. Latitude, longitude, speed, and time
stamps were all included in the GPS data.
The data received in the server was immediately
processed in order to make a real-time prediction of the
bus's arrival time. The time of arrival of the next bus stop
was predicted using data from previous days and real-
time data from a GPS receiver at the current bus stop.
The mobile app enabled for attendance tracking as well
as real bus location tracking, as well as warnings (for
intrusion detection and accident detection). The outputs
for RFID scanning, student detection, and intrusion
detection are depicted in Figures -5, 6 and 7.

Fig -5: RFID Scanningprocesssetup
Fig -6: Student detected on successful scan
Fig -7: Intruder detected on unsuccessful scan

Fig-8: Algorithm model

The implementation algorithm has been broken into
three components, as shown in Fig-8, to make it easier to
understand. The data about the stops and best-estimated
routes is examined and stored on the cloud database
starting at the top. RFID is set up to detect students and
intruders. The image of the invader is captured, saved,
and sent to the cloud. Using a Machine Learning model,
an accelerometer is utilised to calculate and evaluate
driving behaviour for accident detection.
Figure 9 depicts various views of the proposed
implementation of the mobile application. The graphs
below exhibit accelerometer data for accident detection
for several samples (X-Axis – Timeline & Y-Axis – G
Values). The samples were gathered in order to better
understand how people behave in various situations. The
data was then entered into the ML model to make
predictions after it had been wrangled. A 91.8 percent
accuracy rate was obtained.
Fig-9: 3 Apps and a snap enteringbus details Tabs
Fig-10: Live Tracking Tabs
Fig -11: Accident detection and Attendance view
Fig-12a: Sample Data 1
Fig-12b: Sample Data 2

CONCLUSIONS
The prototype was successfully implemented, and the
key functionalities of student detection via RFID,
intrusion detection, location tracking, and accident
detection were all confirmed to be correct. Secondary
capabilities such as attendance analysis, route estimates,
cloud-storage of essential data, and custom alerts were
also tested and found to work as expected. The
prototype outperforms GSM and Arduino-based systems
by combining GPS and Google Maps APIs to give live
location and real-time tracking of children. Overall, this
was a positive step toward a brighter future for Smart
School buses.
REFERENCES
[1] R. Bandal and A. Oak, “Managing Location
Identification and Chain SMS for Smart School Transport
System using IoT,” IEEE 2019 International Conference
on Computer Communication and Informatics (ICCCI -
2019), Jan. 23-25 2019, Coimbatore, INDIA.
[2] B. Pavithra, S. Suchitra, P. Subbalakshmi, J. Mercy
Faustina, “RFID based Smart Automatic Vehicle
Management System for Healthcare Applications,” Third
International Conference on Electronics Communication
and Aerospace Technology [ICECA 2019].
[3] N. Akhtar, K. Pandey, S. Gupta, “Mobile Application
For Safe Driving,” 2014 Fourth International Conference
on Communication Systems and Network Technologies.
[4] J. T. Raj, J. Sankar, “IoT Based Smart School Bus
Monitoring and Notification System,” 2017 IEEE Region
10 Humanitarian Technology Conference (R10-HTC) 21-
23 Dec 2017, Dhaka, Bangladesh.
[5] Jisha R C et al., “An Android Application for School
Bus Tracking and Student Monitoring System,” 2018
IEEE International Conference on Computational
Intelligence and Computing Research.
[6] W. Pattanusorn and I. Nilkhamhang, “Real-Time
Monitoring System for University Buses using Available
Wi-Fi Networks and Travel Time Prediction,” 2018 15th
International Conference on Electrical
Engineering/Electronics, Computer,
Telecommunications and Information Technology.
[7] H. Eren, S. Makinist, E. Akin, and A. Yilmaz, “2012
Intelligent Vehicles Symposium Alcalá de Henares,
Spain, June 3-7, 2012.
[8] M. Ahmad Dar and J. Parvez, “A Live-Tracking
Framework for Smartphones,” IEEE Sponsored 2nd
International Conference on Innovations in Information
Embedded and Communication Systems ICIIECS’15.
Fig-12c: Sample Data 3

A novel method for Smart school bus tracking system using Machine learning and IoT

Recommended

More Related Content

Similar to A novel method for Smart school bus tracking system using Machine learning and IoT (20)

More from IRJET Journal (20)

Recently uploaded (20)

A novel method for Smart school bus tracking system using Machine learning and IoT