A Low-Cost Real-time Road Surface Condition

Monitoring and Alert System

2022 4th International Conference on Emerging Trends in Electrical, Electronic and Communications Engineering (ELECOM) | 978-1-6654-6697-4/22/$31.00 ©2022 IEEE | DOI: 10.1109/ELECOM54934.2022.9965242

Vandana Bassoo Yogesh Beeharry R. Ramjug-Ballgobin

Department of Electrical and Electronic Department of Electrical and Electronic Department of Electrical and Electronic
Engineering Engineering Engineering
University of Mauritius University of Mauritius University of Mauritius
Réduit, Mauritius Réduit, Mauritius Réduit, Mauritius
[email protected] [email protected] [email protected]

Mozammeer Deenmahomed Karan Soin

Department of Electrical and Electronic Pacific Electric Pvt Ltd
Engineering India
University of Mauritius [email protected]
Réduit, Mauritius
[email protected]

Abstract— A good transportation system is essential to wheelers. The device has inbuilt sensors such as a gyroscope,
modern society. Unfortunately, road networks deteriorate accelerometer and GPS and has a module to connect with the
rapidly because of a range of reasons, namely wear and tear 4G network to transfer data in real-time. The data is collected
caused by the continuous and heavy flow of vehicles, weather in real-time on a cloud database and machine learning
elements such as heavy rain and high temperature and various algorithms are used to identify potholes. The spots where
excavation works that are carried out on the roads. Bad roads potholes have been identified are shown on a map. A mobile
lead to accidents, damage to vehicles and increased fuel application is developed to offer an early alert system so that
consumption. Therefore, the monitoring of road quality is road users are notified about potholes on the roads they are
crucial. This paper proposes a low-cost prototype that can be
taking before they reach the location where the road surface is
implemented in any vehicle. The device collects and transmits
data on road quality to a cloud platform, where analytics is
degraded.
performed to assess and classify road condition. The pothole II. LITERATURE REVIEW
locations are viewable on a web page. While travelling, drivers
receive real-time sound alerts on their mobile phones to warn In most cases, accidents result from drivers’ carelessness
them of imminent degradation in the road network, such as involving speeding and unfocused driving. Nevertheless, the
potholes. An Android application has been developed to driver cannot be blamed for many accidents. Poor road
synchronize with the database via the API and send real-time conditions can also cause serious accidents, injuring motorists,
and up-to-date notifications to road users whenever they travellers and pedestrians. Roads conditions can worsen to a
approach such areas. Authorities can also use the system to point where the roads become very unsafe. This includes
grasp the quality of the road network and automatically identify potholes, uneven road surfaces, wrecked concrete, bare
problems that may require immediate repair. reinforcing bar, sinkholes and road fissures. A pothole can be
viewed as a bowl-shaped depression in the road surface [4].
Keywords—pothole detection, decision trees, road surface Driving over a pothole can cause a tire to burst and force the
quality, vibration-based method,
vehicle onto another lane or out of its path, triggering a
I. INTRODUCTION collision with a structure or another vehicle. Also, driving on
uneven roads increases the risk of the driver losing control of
Poor road conditions lead to discomfort during travel, his vehicle and crashing or rolling over, resulting in serious
damage to vehicles and tragic accidents. A 2014 study by the injuries. The injuries caused by accidents include head trauma,
American Automobile Association reported that the bill to fix brain injury, neck/spinal injuries, bone fractures, eye injuries
damaged vehicles due to bad roads amounted to 6.4 billion and internal injuries [5]. Some injuries can cause a permanent
dollars [1]. The most common damages were punctured tyres, reduction in mobility and may require costly lifetime
bent wheels and broken suspension. It was also reported that treatment.
road-related conditions caused one-third of all traffic fatalities
in the US. In a 2016 article, the Economist reported that 3,000 A large amount of information related to road traffic is
people are killed every year in India in accidents caused by obtained via traffic monitoring systems [6] and vehicles’ GPS
potholes [2]. Delayed repairs are also more expensive. It was systems [7], [8]. However, humans are still collecting
reported that repair costs of roads increase by 7 times if the important information and this includes road surface
repairs are carried out five years later [1]. Bad road problems monitoring. As mentioned previously, road surface state
afflict poor and rich countries alike and Mauritius is no defines road quality which ensures efficient fuel consumption
exception. Our roads suffer from similar problems. Damages as well as the safety of road users. The conception of an
to wheel rims are common and many insurance policies do not automated system for road surface monitoring is quite
cover such accidents [3]. complex and would be of great benefit to the public.
This paper proposes a road surface monitoring system that Yu and Yu [9] used new data acquisition hardware to
uses a low-cost device that can be installed on any four conceive a system that evaluates road conditions through
vibrations. The authors recognized the impact of cracks and rutting on

978-1-6654-6697-4/22/$31.00 ©2022 IEEE

Authorized licensed use limited to: Jain University. Downloaded on February 12,2025 at 10:44:15 UTC from IEEE Xplore. Restrictions apply.
vehicles. The conditions of the surface are assessed based on Sensing While While going While While
the saved data from a test vehicle driving on the road. The period approachi over a approaching approachi
ng pothole ng
device does not require large space for storage, is cost-
effective and can be applied to real-time data processing. Processing High - Low – High – High –
However, the system does not provide the details of the Complex Readings Collections of Complex
image are used 3D points 3D image
problems on the road surface. Moreover, the vehicle's processin directly processing
condition must be calibrated to compare the results. g

With the wide use of the internet, most machine to Cost High Low High High
machine and human to machine interactions are being Detection Difficult Can detect Can detect Difficult
generated by the transmission and reception of digital signals at night
via the Internet of Things (IoT). This theory is very useful in
Accuracy Depends High High Depends
the emission of a large amount of information which can on on camera
eventually help in the conception of helpful systems in the algorithm alignment
home and industrial automation [6] . IoT technologies are s used s &
recognized as a means of optimization for various public algorithms
used
facilities such as transportation, street lighting, observation
and protection of public zones among others [10].
Recently, there has been a lot of research interest in III. METHODOLOGY
devising methods to identify road problems, classify roads
according to its quality and alert drivers of impending This research project aims to develop a low-cost device
problems. The authors of [11] studied and classified existing that can be installed on any four wheelers vehicle to collect
pothole detection techniques into three categories: vibration- data from different sensors, which will be used to assess road
based methods, 3D reconstruction-based methods, and vision- quality and provide alerts to road users and information to the
based methods. The authors reported that, even though the relevant authorities.
vision-based methods are more cost-effective than the 3D The objectives of this work are:
laser-scanner methods, the signals that they generate can be
easily affected by noise. This is because they perform • To design and produce a low-cost device to capture
detection by considering images and videos. Therefore, it is data to assess road quality in Mauritius.
necessary to design a method that uses several 2D data to • To implement a cloud-based platform to store the data
enhance the current methods and improve their accuracies. An collected in real-time from moving vehicles.
overview of the different methods proposed for detecting
potholes by vibration-based, vision-based, laser-based and • To employ classification-based machine learning
stereo imaging methods are shown in Table I. algorithms to categorize road quality.
Jaguar Land Rover has developed a system where their • To develop a mobile application to alert drivers of the
vehicles gather information about potholes, broken drains and degradation of road quality.
manhole covers [12]. The information is relayed to other
drivers allowing them to slow down or adjust their suspension A. Hardware
when they approach those problem areas. However, this The hardware consists of the following components:
system is embedded in the vehicle and is accessible only to
Range Rover Evoque owners. • Sensor Module (IMU),

The authors in [13] proposed algorithms to detect potholes • Processing Unit (Microcontroller),
using accelerometers found in Android smartphones. The data • Data Storage Unit (µSD card),
was collected and processed offline. The authors in [14]
developed a similar system using accelerometer and GPS data. • Communication Unit (SIM7600EI),
Machine learning algorithms were applied to identify road
• GPS Unit L80R (QUECTEL),
surface anomalies. Given that the accelerometer data gives
information only on linear movement, it is not sufficient to use • Power Management and Battery (MIC5219),
the accelerometer alone for accurate detection of road quality.
• User Interface consisting of:
TABLE I. DETECTION OF POTHOLES USING FOUR DIFFERENT
METHODS [15]
o OLED Display,
Vision- Vibration- Laser-based Stereo o Buttons,
based based Imaging
o LED Indicators.
Devices Camera Sensors Laser Cameras
The sensor module contains the MPU 6050 chip and
Technolog 2D Rotation, 3D 3D supporting passive components. The MPU 6050 is the inertial
y used Imaging Speed, reconstruction reconstruc
Orientation of image using tion using measurement unit (IMU), which uses the MEMS technology
measured light reflection multiple to measure vibration's linear and circular components of
by cameras vibration [16]. This chip consists of a three-axis accelerometer
gyroscope and a three-axis gyroscope, along with a temperature sensor to
&
acceleromet keep track of any subtle offsets in the measurements. The
er MPU6050 IMU can sample data from 50 samples per second
(SPS) to 1000 samples per second (user configurable) (we are

Authorized licensed use limited to: Jain University. Downloaded on February 12,2025 at 10:44:15 UTC from IEEE Xplore. Restrictions apply.
using 100 SPS setting). Mobile phones with in-built
gyroscope and accelerometer cannot provide large SPS,
limiting the accuracy of systems. Therefore, a custom-made
device with a specialist chip has been chosen for this project.
This chip uses an inter-integrated circuit (I2C) bus to
communicate with the host (microcontroller).
The processing unit for this prototype is a 16-bit PIC
microcontroller device namely PIC24FJ256GB106. The
microcontroller has several communication peripheral Fig. 1. The controller board
modules like the I2C bus, serial peripheral interface (SPI) bus
and UARTs (Microchip, 2020) , which we use to
communicate to several blocks in our prototype. A microSD
card is used to store the data generated by the sensor module.
The microcontroller communicates with the microSD using
the SPI bus.
For the communication and data transmission from the
prototype to the cloud API, the SIM7600EI 4G
communication module is used in the prototype. The
SIM7600EI device is a complete multi-band LTE-FDD/LTE-
TDD/HSPA+/UMTS/EDGE/GPRS/GSM module solution in Fig. 2. The communication board
LCC type and it supports LTE CAT1 up to 5 Mbps for uplink
and 10 Mbps for downlink data transfer. The module is
compatible with SIMCom HSPA+ SIM5360 module/LTE
CAT3 SIM7100 and LTE CAT4 SIM7600E-H module and
allows customers to design their application once for different
technology and benefit from great development timesaving.
The module has a wide range of application capabilities such
The User Interface
as TCP/UDP/FTP/FTPS/HTTP/HTTPS/SMTP/POP3 and
MMS which provide a lot of flexibility and can easily be
integrated with third-party applications [17]. The GPS Unit
The GPS unit QUECTEL L80-R is a patch antenna on top
(POT) GPS module. The Embedded Assist System for self- The Data Storage
generated orbit prediction, EASY™ technology assist the Unit
L80-R for calculating and predicting orbits automatically by Fig. 3. The Testing Jig
using up to 3 days of ephemeris data stored in the RAM, so it
can fix position quickly and with low power consumption,
even when signal levels are lower [18]. The module’s ultra- B. Software
compact design, low cost and low power demand along with
its high sensitivity and precision makes it ideal for a wide The main blocks for the software part are shown in Fig. 4
range of IoT applications such as portable devices, and consist of:
automotive, security, personal tracking and industrial PDAs. 1. The firmware of the device is placed in vehicles to
The module also suits special applications such as GPS mouse collect and transfer data to the online API,
receivers and onboard diagnostics [18]. 2. A REST API,
The unit is powered with an onboard 3.7-volt Li-ion 3. A cloud database,
battery. The battery in the unit is charged using an external 4. A web interface / Dashboard,
charger which is powered via USB. This can be done by using 5. An Android mobile application.
the car’s cigarette lighter port or alternatively, a power bank
can be used. A 5V system was chosen over 12V to make the
circuit more efficient and more compact. The output from the
battery is fed into a Low Dropout regulator (MIC5219) to
provide a stable 3.3 Volt supply for the circuit.
The user interface consists of an OLED display, two
micro-tactile switches and two LEDs to allow the user to view
and configure basic settings, data and statuses.
The prototype was implemented, and photos of the circuit
boards and casing shown in Fig. 1-3.

Fig. 4. The main blocks for the software part

Authorized licensed use limited to: Jain University. Downloaded on February 12,2025 at 10:44:15 UTC from IEEE Xplore. Restrictions apply.
 The firmware is written in C programming language, more than 288 000 rows was leveraged to train the machine
mainly for sensor data collection and transfer. The REST API learning model to ensure high accuracy.
acts as an interface to transfer data to and from the cloud
database. It also uses trained machine learning models to Data collected by the prototype was successfully
detect potholes and humps from the data received from the transferred via 4G network to the REST API. The collected
hardware and the road surface type detected is saved in the data is also stored on an SD card and can be uploaded via the
cloud database along with each set of data received. The REST web interface. The REST API leveraged the machine learning
API is also connected to an online web interface and also algorithms to classify the data successfully and inserted into
serves a client Android application. The machine learning the online database. Some of the potholes identified by the
algorithm library namely scikit-learn has been used in the machine learning model are shown in Fig. 8.
Python programming language for this project as it offers a
wide variety of machine learning algorithms which can be
used for classification and regression [19]. The models used
are: Quadratic Discriminant Analysis, AdaBoost classifier,
Naive Bayes, Random Forest, Linear Discriminant Analysis,
Gradient Boosting and Decision Trees. The web interface
comprises of data filtering, data upload and download and
Google Map features. The cloud database system used is
ElephantSQL since it is SQL compliant and an open-source
objective Relational Database Management System
(RDBMS) [20]. Finally, the android application has
geofencing, location update and retrofit services integrated for
proper real-time notifications and location updates.
C. Testing
The device was fitted with a sim card and operated. A few
interfaces are shown in Fig. 5-7.
Fig. 8. Some of the identified potholes by the ML model through data
collected from the prototype

IV. CONCLUSION
This research aimed at developing a prototype to assess
road quality and warn drivers of degradation in the quality of
the road. The project was ambitious, complex, and multi-
disciplinary. It entailed the design and manufacture of an
electronic device, using a cloud platform and machine
Fig. 5. Display of accelerometer and gyroscope readings learning algorithms, and developing a web page and a mobile
application.
A fully operational prototype that can perform road
surface monitoring and alert road users about poor road
surface conditions was developed. The device has
successfully transferred real-time data collected from its in-
built sensors to a cloud-based REST API. The REST API used
machine learning algorithms to classify the road surface
condition. Potholes detected are displayed on a map and a
Fig. 6. Display of GPS information mobile application has been developed to notify road users in
real-time as they approach potholes.
Authorities can also potentially use the system to
understand the quality of the road network and identify
problems that may require immediate repair.
ACKNOWLEDGMENT
The project was funded by the University of Mauritius
through the RFS-A grant - RFS-A2-19/20.
Fig. 7. Network Status
REFERENCES
The prototype was placed on the dashboard and data was
collected. Data collected from the prototype was used to train
the machine learning model. Two methods were used to [1] K. Wiggers, "Google patents a system for cars that maps
identify potholes for training purposes. Initially, dash cam potholes," 2015. [Online]. Available:
video footage was used but it was found to be not practical for https://www.digitaltrends.com/android/google-pothole-
long drives. A second method was adopted. A GPS marker detection/. [Accessed 7 4 2019].
mobile application was used to mark pothole areas while [2] The Economist, "The hole story," The Economist, 11 June 2016.
collecting data with the prototype. A dataset consisting of

Authorized licensed use limited to: Jain University. Downloaded on February 12,2025 at 10:44:15 UTC from IEEE Xplore. Restrictions apply.
[3] Le Défi Quotidien, "Quelle est l’utilité de payer la road tax?," Smartphones with Accelerometers," in 7th IEEE International
2019. [Online]. Available: https://m.defimedia.info/courrier-de- Conference on Distributed Computing in Sensor Systems and
lecteur-quelle-est-lutilite-de-payer-la-road-tax. [Accessed 7 4 Workshops , Barcelona, 2011.
2019]. [14] J. Eriksson, G. Lewis , B. Hull, R. Newton, S. Madden and H.
[4] J. S. Miller and W. Y. Bellinger, "Distress identification manual Balakrishnan, "The Pothole Patrol: Using a Mobile Sensor
for the long-term pavement performance program," United Network for Road Surface Monitoring," in MobiSys, Colorado,
States. Federal Highway Administration. Office of Infrastructure 2008.
Research and Development, United States, 2003. [15] A. A. Shaghouri, R. Alkhatib and S. Berjaoui, "Real-time
[5] Gilman and Bedigian, "Accidents caused by poor road quality pothole detection using deep learning," arXiv preprint
and Conditions," [Online]. Available: arXiv:2107.06356, 2021.
https://www.gilmanbedigian.com/accidents-caused-by-poor- [16] TDK InvenSense, "MPU-6050 Six-Axis (Gyro + Accelerometer)
road-quality-and-conditions-in- MEMS MotionTracking™ Device," 2020. [Online]. Available:
dc#:~:text=Accidents%20Caused%20by%20Poor%20Road%2 https://invensense.tdk.com/products/motion-tracking/6-
0Conditions,than%20just%20a%20bumpy%20ride.&text=Unev axis/mpu-6050/. [Accessed 20 October 2020].
en%20road%20surfaces%20can%20cause,driver%2C%20passe
ngers%2C%20an. [Accessed 11 June 2020]. [17] SIMCom, "SIM7600EI 4G LTE High-Speed Modem
GPS/GNSS IoT board Raspberry Pi Compatible," 2020.
[6] A. Zanella, N. Bui, A. Castellani, L. Vangelista and M. Zorzi, [Online]. Available:
"Internet of things for smart cities," IEEE Internet of things https://www.electronicscomp.com/sim7600ei-4g-lte-high-
journal, pp. 22-32, 2014. speed-modem-gps-gnss-iot-board-raspberry-pi-
[7] X. Li, W. Shu, M. Li, H.-Y. Huang, P.-E. Luo and M.-Y. Wu, compatible#:~:text=SIM7600EI%20is%20a%20complete%20
"Performance evaluation of vehicle-based mobile sensor multi,5Mbps%20for%20uplink%20data%20transfer..
networks for traffic monitoring," IEEE transactions on vehicular [Accessed 20 October 2020].
technology, pp. 1647-1653, 2008. [18] Quectel, "Quectel L80-R Compact GPS Module Integrated with
[8] T. H. Silva, P. O. V. De Melo, A. C. Viana, J. Sales and A. A. F. Patch Antenna," 2017. [Online]. Available:
Loureiro, "Traffic condition is more than colored lines on a map: https://www.quectel.com/wp-
characterization of waze alerts," in International Conference on content/uploads/pdfupload/Quectel_L80-
Social Informatics, WA, USA, 2013. R_GPS_Specification_V1.2.pdf. [Accessed 24 May 2021].
[9] B. X. Yu and X. Yu, "Vibration-based system for pavement [19] S. Pierre, "A COMPREHENSIVE GUIDE TO SCIKIT-LEARN
condition evaluation," Applications of Advanced Technology in (SKLEARN)," 2021. [Online]. Available:
Transportation, pp. 183-189, 2006. https://builtin.com/machine-learning/scikit-learn-guide.
[10] M. Dohler, I. Vilajosana, J. Llosa and X. Vilajosana, "Smart [Accessed 3 March 2021].
Cities: An action plan," in Barcelona Smart Cities Congress, [20] PostgreSQL, "PostgreSQL: The World's Most Advanced Open
Barcelona, Spain, 2011. Source Relational Database," 2020. [Online]. Available:
[11] T. Kim and S.-K. Ryu, "Review and analysis of pothole detection https://www.postgresql.org/. [Accessed 12 November 2020].
methods," Journal of Emerging Trends in Computing and [21] Microchip, "PIC24FJ256GB106 General Purpose USB
Information Sciences, vol. 5, no. 8, pp. 603-608, 2014. Microcontroller," 2020. [Online]. Available:
[12] Jaguar Land Rover, "Pothole Detection Technology Research https://www.microchip.com/wwwproducts/en/PIC24FJ256GB1
Announced By Jaguar Land Rover," 2015. [Online]. Available: 06#additional-features. [Accessed 20 October 2020].
https://www.landrover.com/experiences/news/pothole-
detection.html. [Accessed 7 4 2019].
[13] A. Mednis, G. Strazdins, Z. Reinholds , K. Georgijs and L.
Selavo, "Real Time Pothole Detection using Android

Authorized licensed use limited to: Jain University. Downloaded on February 12,2025 at 10:44:15 UTC from IEEE Xplore. Restrictions apply.