12 IV April 2024

https://doi.org/10.22214/ijraset.2024.60686
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at www.ijraset.com

Real-Time Traffic Light Optimization Using AI

and IOT
Aniket Phand1, Shweta Bagade2, Nikhil Bandgar3, Prof. Ganesh Wayal4
Department of Computer Engineering, Padmabhooshan Vasantdada Institute of Technology, Pune, India

Abstract: Urban traffic congestion is a significant challenge globally, impacting transportation efficiency, environmental
sustainability, and urban liveability. Traditional traffic control systems often struggle to adapt to changing traffic dynamics,
leading to increased congestion and delays. This paper presents a novel traffic management system developed by our team,
leveraging cutting-edge technologies such as computer vision, artificial intelligence (AI), and the Internet of Things (IoT).
Deployed at intersections, our system utilizes real-time CCTV feeds for traffic analysis, employing advanced image processing
and machine learning algorithms, including YOLO V7, to dynamically assess traffic density. These insights inform adaptive
adjustments to traffic light timings, with the aim of reducing congestion, enhancing transit efficiency, and mitigating pollution.
Our signal switching algorithm, incorporating dynamic logic developed by our team, iteratively adjusts signal timings based on
real-time traffic conditions, ensuring responsive and efficient traffic flow management. Through the integration of our
innovative technologies, our system seeks to revolutionize urban transportation networks, promoting smarter, more adaptive, and
sustainable urban environments.
Keywords: Traffic Management, Urban Congestion, Intelligent Systems, Computer Vision, Artificial Intelligence, Machine
Learning, Deep Learning, Traffic Prediction, Adaptive Control, Smart Cities, YOLO.

I. INTRODUCTION
In the ever-growing cities, the rise in the number of vehicles has caused problems on roads, leading to reduced capacity and a drop
in service quality. One critical issue contributing to this is the old-fashioned traffic control systems at intersections, using fixed
signal timers. These systems repeat the same patterns without adapting to the changing traffic conditions, causing more traffic and
inefficiency. To tackle these challenges, this project explores a new way of controlling traffic. It discusses the drawbacks of manual
control, fixed-timer traffic lights, and electronic sensors, paving the way for an innovative solution using Computer Vision. By
using live CCTV images at intersections, this system aims to figure out real-time traffic density, categorize vehicles, and adjust
signal timings dynamically. This approach promises a more responsive, efficient, and eco-friendly traffic management system.
Develop an intelligent traffic light management system using Computer Vision, Artificial Intelligence (AI), and Internet of Things
(IoT) to achieve real-time traffic density calculation, vehicle classification, and adaptive signal timing. The primary goal is to reduce
congestion, enhance traffic flow, and minimize environmental impact for a more efficient and sustainable urban transportation The
project aims to ease congestion at intersections, reducing delays and pollution caused by the increasing number of vehicles in urban
areas. Motivated by the limitations of current traffic control systems, it seeks to introduce an intelligent, responsive system using
advanced technologies like Computer Vision and AI. This approach, beyond conventional methods, adapts traffic signals
dynamically based on real-time conditions. With case studies from cities like Mumbai and Bangalore emphasizing the need for
innovation, the project aims to enhance traffic control effectiveness, reduce travel times, and contribute to a sustainable urban
transportation system. By integrating IoT devices, the system will gather real-time data on traffic flow, enabling precise adjustments
to optimize signal timings further. Additionally, the project will explore AI algorithms to predict traffic patterns, facilitating
proactive measures for congestion management. Through continuous monitoring and adaptation, the proposed system aims to
revolutionize urban traffic management, offering a scalable and sustainable solution for future cities.

II. BACKGROUND
We get motivated by disadvantages of the existing system. The purpose of the system is to decrease congestion at intersections
primarily, and reduce delays and pollution. The motivation to commence on this project arises from the critical issues linked to the
surging number of vehicles in urban areas, leading to worsening traffic congestion. This congestion not only causes delays and
stress for drivers but also contributes to higher fuel consumption and increased air pollution.

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3534
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at www.ijraset.com

The negative impacts are particularly evident in megacities, underscoring the urgency of addressing the challenges of traffic
management in real-time. The current traffic control systems, often reliant on fixed signal timers and outdated methods, prove
insufficient in adapting to the dynamic nature of urban traffic. This project is motivated by the need to revolutionize traffic
management and introduce a more intelligent and responsive system that can mitigate congestion, improve transit efficiency, and
contribute to a more sustainable and eco-friendlier urban environment. Cities like Mumbai and Bangalore serve as impactful case
studies, highlighting the urgency of finding innovative solutions. The excessively long travel times during rush hours in these cities
underscore the inadequacies of traditional traffic control methods. The motivation is to create a system that goes beyond
conventional approaches, leveraging a suite of cutting-edge technologies such as Computer Vision, Artificial Intelligence (AI), and
Internet of Things (IoT) to dynamically adjust traffic signals based on real-time conditions.

III. LITERATURE REVIEW

1) Paper A : Smart Control of Traffic Light Using Artificial Intelligence:
This paper addresses the escalating issue of traffic congestion in cities, exacerbated by population growth and increased automobile
usage. Traffic jams not only cause delays and stress for drivers but also lead to higher fuel consumption and air pollution,
particularly affecting megacities. Real-time calculation of road traffic density is crucial for effective signal control and traffic
management. The proposed system utilizes live images from traffic junction cameras for traffic density calculation through image
processing and AI. Additionally, it presents an algorithm for adaptive traffic light control based on vehicle density to mitigate
congestion, ultimately aiming to provide faster transit and reduce pollution.[2]
Limitations:
 Reliance on live camera feeds may introduce delays and inaccuracies due to environmental factors.
 Continuous calibration is required for optimal performance, which may be resource-intensive, time-consuming.
Metrics Their Approach Our Approach
Traffic Flow Efficiency Traffic flow efficiency various methods such as Implementation of AI
traffic light optimization, dynamic signal control. prediction models and
dynamic signal control.
Congestion Reduction Relies on traffic light optimization, road network Integrates AI driven
modification. congestion prediction and
dynamic traffic rerouting.
Dynamic Routing Limited or absent traditional routing algorithms Integrates dynamic routing
are used. based on real time traffic
conditions and congestion
predictions.
Signal Switching Algorithms Traditional algorithms like fixed time or actuated Incorporates adaptive signal
signal control. control algorithms using
reinforcement learning or
evolutionary strategies.
Object Detection Algorithms YOLO V3: Real-time object detection algorithm. YOLO V7 is a real-time
object detection algorithm,
more optimized.

2) Paper B : Traffic Flow Prediction for Smart Traffic Lights Using Machine Learning Algorithms:
The paper focuses on predicting traffic flow at intersections using machine learning and deep learning algorithms to facilitate
adaptive traffic control. Utilizing public datasets, ML and DL models are trained and validated for traffic flow prediction, laying the
groundwork for smart traffic light controllers. While the proposed models show promising performance, limitations may include the
need for extensive data preprocessing and model training as well as potential challenges in real world deployment due to variations
in traffic patterns and environmental factors.[4] .
Limitations:
 Extensive data preprocessing and model training are needed, which may require significant computational resources.

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3535
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at www.ijraset.com

 Real world deployment may face challenges due to variations in traffic patterns and environmental conditions.
Metrics Their Approach Our Approach

Traffic Flow Efficiency Traffic Flow Efficiency Utilizes Implement Real-time traffic flow
Historical Traffic Data for prediction using machine learning
prediction. algorithms

Congestion Reduction Relies On Historical Data analysis Utilizes Machine Learning models
for congestion prediction. for real-time congestion prediction
Linear Regression and Stochastic and adaptive signal control
Gradient Regressor is used to
predict traffic flow.

Dynamic Routing Absent Incorporates Dynamic Routing

based on real-time traffic
predictions Signal Switching
Algorithms

Signal Switching Algorithms Not specified Utilizes Machine Learning-based

adaptive signal

Object Detection Algorithms Not specified YOLO V7 is a real-time object

detection algorithm, more
optimized

Number Of Stops Uses historical data analysis to Employs Machine Learning models
optimize signal timings for real-time congestion monitoring
and proactive signal control to
minimize stops.

3) Paper C : Design and Development of Portable Smart Traffic Signalling System with Cloud Artificial Intelligence
Enablement:
The paper introduces a cost-effective, IoT-enabled portable traffic signalling system designed for smaller traffic junctions. Utilizing
ESP32 microcontrollers and foldable mechanical structures, the system offers manual and automatic traffic control modes, with the
ability to gather traffic density information and optimize signal timings based on cloud-based algorithms. While the system
demonstrates robustness and accuracy in traffic management, limitations may include scalability issues for larger intersections and
dependence on constable internet connectivity for cloud-based operations.[1] .
Limitations:
 Scalability Issues May Arise for larger intersections, limiting the system's applicability.
 Dependence Unstable Internet Connectivity for Cloud-based operations may hinder functionality in areas with poor
connectivity.
Metrics Their Approach Our Approach

Traffic Flow Efficiency Traffic Flow Efficiency Relies Implements Smart Traffic Signal
Fixed signal timings or manual control based on real-time traffic
control. data and cloud-based AI algorithms.

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3536
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at www.ijraset.com

Congestion Reduction Limited Ability to Adapt to Utilizes cloud-based AI algorithms

Changing Traffic conditions. to analyse traffic data and adjust
signal timings dynamically to
reduce congestion.

Dynamic Routing Routing Decisions Are Not Incorporates Dynamic Routing

influenced by real-time traffic based on real-time traffic prediction
conditions. s and cloud-based AI analysis.

Signal Switching Algorithms K-means clustering for obtaining Uses vehicle density, dynamic
optimized time values to switch algorithm, to set the green signal
timer for each lane.

Object Detection Algorithms Absent YOLO V7 is a real-time object

detection algorithm, more
optimized.

4) Paper D : Real-Time Traffic Light Optimization Using Simulation of Urban Mobility:

The paper presents a study on real-time traffic light optimization using the SUMO tool for simulating urban mobility. Testing
Scenarios in Berlin city, the study analyses traffic performance under different scenarios, optimizing traffic flow by modifying
various parameters such as traffic lights and road construction. While the study demonstrates the potential for optimizing traffic
performance, limitations may include the complexity of accurately modelling real-world traffic behaviour and the challenge of
extrapolating simulation results to practical implementations diverse urban environments.[5]
Limitations:
 Accurately modelling real-world traffic behaviour simulation may be complex and challenging.
 Extrapolating simulation results to practical implementations in diverse environments may not fully capture real world
complexities.
Metrics Their Approach Our Approach

Traffic Congestion Mitigation Utilizes SUMO tool for traffic Utilizes Computer Vision, and DL
simulation and optimization algorithms for dynamic traffic
light adjustments.

Transit Efficiency Focuses Traffic Light optimization but Provides real time traffic info,
Improvement lacks detailed discussion transit optimal routes and estimated
efficiency improvement for drivers. transit times through a mobile app.

Environmental Impact Aims to reduce pollution through Aims to Minimize Pollution By

Reduction traffic optimization but lacks explicit reducing congestion and
discussion on environmental impact. optimizing traffic flow.

Dynamic and Responsive Identifies limitations of current traffic Integration of YOLO V5 with
Traffic Management control systems but does not delve dynamic signal switching allows
into specific critiques of traditional for responsive and adaptive
methods. Dijkstra’s algorithm has systems that can react to changes
been chosen to explore the edges in in their environment in real-time
the traffic scenario. based on detected objects , events.

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3537
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at www.ijraset.com

IV. COMPARISON OF ALGORITHMS EFFECTIVENESS

Metrics Paper A[2] Paper B[4] Paper C[1] Paper D[5]

Algorithm Adaptive traffic Machine learning K-means clustering Simulation of Urban

light control based and deep learning for generating Mobility (SUMO)
on real-time vehicle algorithms for optimized time tool for real-time
density using image predicting traffic values based on traffic light
processing and AI. flow at intersections. traffic density data. optimization.

Effectiveness Demonstrated a Multilayer Demonstrated Optimized traffic

significant 23% Perceptron Neural robustness and performance by
improvement over Network (MLP-NN) accuracy in wireless modifying various
the current system in achieved the best data communication traffic scenario
terms of the number performance with an and traffic factors such as
of vehicles crossing R-squared score of management. traffic lights and
the intersection. 0.93 junctions.

Speed Real-time Offline training time Fast computation of Simulation speed

processing of traffic varies depending on cluster centroids, depends on the
camera feeds, with the dataset size and typically taking complexity of the
an average complexity of the milliseconds to traffic scenario and
processing time of model architecture, compute. Real-time computational
100 milliseconds per typically ranging adjustment of traffic resources. Real-time
frame. from several hours signal timings with simulation
to a few days. minimal latency. achievable with fast
computational
hardware (multi -
core GPUs).

V. CONCLUSIONS
The proposed intelligent traffic management system presents a promising solution to alleviate urban traffic congestion and enhance
transit efficiency.
Through the utilization of computer vision and AI technologies, the system endeavours to overcome the shortcomings of
conventional traffic control systems, fostering the creation of a smarter and more efficient urban transportation network.
Nonetheless, challenges pertaining to data accuracy and scalability persist, underscoring the necessity for continuous research and
development in this domain.
Moreover, the system's ability to minimize unnecessary stops and starts not only improves traffic flow but also results in substantial
fuel savings and emissions reduction, thereby offering a positive environmental impact. Moving forward, further advancements in
algorithm optimization, real-time data processing, and seamless integration with existing infrastructure are imperative to realize the
full potential of intelligent traffic management systems in urban environments.

VI. FUTURE DIRECTIONS

Future research directions include further optimization of AI algorithms for traffic prediction and control, integration of IoT
technologies for enhanced data collection, and collaboration with city authorities for real-world implementation and testing.
Additionally, exploring hybrid approaches that combine the strengths of different algorithms could yield more adaptable and
efficient solutions. Moreover, integrating emerging technologies like edge computing and 5G connectivity could further enhance
real-time data processing and decision-making capabilities for smarter urban mobility systems.

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3538
 International Journal for Research in Applied Science & Engineering Technology (IJRASET)
ISSN: 2321-9653; IC Value: 45.98; SJ Impact Factor: 7.538
Volume 12 Issue IV Apr 2024- Available at www.ijraset.com

REFERENCES
[1] Design and development of portable smart traffic signalling system with cloud-artificial intelligence enablement . Indonesian Journal of Electrical Engineering
and Computer Science Vol. 26, No. 1, April 2022, pp. 116~126 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v26.i1.pp116-126 . Badarinath Kamisetty 1 , Mahesh
Renduchintala1 , Lochan Lingaraja Shetty2 , Suresh Chandarshekar2 , Rajashree Shettar11 .
[2] Smart Control of Traffic Light Using Artificial Intelligence. 78-1-7281-8867-6/20/$31.00 ©2020 IEEEDOI:10.1109/ ICRAIE 51050.2020. 9358334. Mihir M.
Gandhi, Devansh S. Solanki, Rutwij S. Daptardar, Nirmala Shinde Baloorkar.
[3] Lilhore, U.K.; Imoize, A.L.; Li, C.-T.; Simaiya, S.; Pani, S.K.; Goyal, N.; Kumar, A.; Lee, C.-C. Design and Implementation of an ML and IoT Based Adaptive
Traffic-Management System for Smart Cities. Sensors 2022, 22, 2908. https://doi.org/10.3390/s22082908
[4] Navarro-Espinoza, A.; López-Bonilla, O.R.; García-Guerrero, E.E.; Tlelo-Cuautle, E.; López-Mancilla, D.; Hernández-Mejía, C.; Inzunza-González, E. Traffic
Flow Prediction for Smart Traffic Lights Using Machine Learning Algorithms. Technologies 2022, 10, 5. https://doi.org/10.3390/technologies10010005.
[5] .Real Time Traffic Light Optimization Using Simulation of Urban Mobility Tanya Garg1·Gurjinder Kaur· Prashant Singh Rana2SN Computer Science (2023)
4:526 https://doi.org/10.1007/s42979-023-01916-9 .
[6] ‘Pygame Library’, 2019. [Online]. Available: https://www.pygame.org/wiki/about.

©IJRASET: All Rights are Reserved | SJ Impact Factor 7.538 | ISRA Journal Impact Factor 7.894 | 3539