AI-Driven Traffic Light Controller Based on Vehicle Density Analysis

Abstract

Managing urban traffic efficiently has become a pressing challenge due to the increasing number of vehicles on the road. Traditional fixed-timing traffic systems fail to adapt to fluctuating traffic conditions, resulting in prolonged delays, unnecessary stops, and heightened environmental impact. Manual interventions by traffic personnel add to the inefficiency, demanding significant manpower while remaining static and error-prone. This project proposes a smart, AI-driven traffic management system which uses the YOLO model for real-time vehicle detection and density analysis. By dynamically adjusting signal durations based on vehicle flow, the system ensures optimal traffic movement, prioritizing heavily congested lanes. This adaptive approach reduces wait times, minimizes fuel consumption, and lowers emissions, addressing both commuter convenience and environmental sustainability. The system operates autonomously, eliminating human errors and reducing the need for manual control. It promotes smoother traffic flow, reduces operational costs, and improves safety for all road users. With its ability to respond to real-time data, this AI-powered traffic controller provides a transformative solution to urban traffic challenges, contributing to a more efficient and eco-friendly transportation ecosystem.

Introduction

Urban traffic congestion is a major problem caused by population growth and increased vehicles, leading to delays, fuel waste, pollution, and stress. Traditional fixed-timing traffic signals and manual traffic police management are inefficient and inflexible, causing unnecessary waiting and environmental harm.

To address this, the project proposes an AI-driven traffic light controller using the YOLO model for real-time vehicle detection and density analysis. The system dynamically adjusts traffic signal timings based on actual vehicle counts, prioritizing congested lanes to optimize flow, reduce waiting times, fuel consumption, and emissions. This automation reduces human error and improves road safety and commuter convenience.

Objectives include developing a deep learning-based adaptive system that improves traffic flow by reducing congestion and demonstrates the effectiveness of AI in urban traffic management.

The literature review highlights recent advances integrating machine learning, image processing, and YOLO for smart traffic control, showing improved adaptability over traditional systems.

Methodology involves capturing images from intersections, using YOLO and TensorFlow to detect and count vehicles, and adjusting signal timings based on traffic flow levels (high, medium, low). The system factors in saturation flow and amber times to optimize green light duration, logs data for future improvements, and processes video frames through enhancement and normalization techniques before detection.

The system works by dividing input images into grids, predicting bounding boxes with confidence scores for detected vehicles, and applying non-max suppression to finalize detections.

Results and discussions demonstrate that the intelligent system enhances traffic signal control through dynamic adjustments, image processing, and data storage, contributing to more efficient and sustainable urban traffic management.

Conclusion

Managing urban traffic has become an increasingly difficult task due to the rapid rise in vehicle numbers on the roads. Traditional traffic signal systems that operate on fixed timings fail to adapt to real-time traffic conditions, leading to significant issues such as longer wait times, unnecessary fuel consumption, increased air pollution, and elevated stress for drivers. Additionally, manual traffic management by personnel is prone to errors and demands a considerable workforce. To address these inefficiencies, this project introduces an AI-driven traffic light controller that utilizes the YOLO (You Only Look Once) model for real-time vehicle detection and density analysis. The system dynamically adjusts signal durations based on the actual number of vehicles at an intersection, ensuring smoother and faster traffic movement while prioritizing heavily congested lanes. The proposed system offers several advantages over traditional methods. By reducing traffic congestion through adaptive signal control, it minimizes delays and vehicle idling at intersections, thereby enhancing the overall flow of traffic. This also contributes to environmental sustainability by cutting down on fuel consumption and reducing greenhouse gas emissions, which helps improve urban air quality. Moreover, the system enhances the commuting experience by saving drivers\' time, lowering frustration, and boosting productivity. With automated traffic management, it eliminates the potential for human error, promoting safer road usage for everyone. The system uses advanced AI tools such as TensorFlow and OpenCV to process images, detect vehicles in real-time, and optimize signal timings. Factors like traffic flow rates and saturation flow are analysed to calculate the appropriate green light durations, ensuring that the system adjusts dynamically to changing traffic conditions. Additionally, it stores and logs traffic data for future analysis, enabling continuous improvement in traffic planning and management, particularly during peak hours. This adaptive approach makes the system highly responsive and efficient. This project represents a significant step toward smart city infrastructure, offering a scalable and transformative solution to urban traffic challenges. By integrating cutting-edge AI techniques, the system not only resolves immediate traffic issues but also supports long-term goals such as sustainability, efficiency, and safety. In doing so, it lays a solid foundation for future advancements in intelligent transportation systems, helping create eco-friendly, efficient, and commuter-friendly urban environments

References

[1] Dr.A. Ravi, R.Nandhini, K.Bhuvaneshwari, J.Divya, K.Janani, “Traffic Management System using Machine Learning Algorithm”,International Journal of Innovative Research in Technology,Volume 7, Issue 11,April 2021. [2] Nitin Sakhare, Mrunal Hedau, Gokul B. Omkar Malpure, Trupti Shah, Anup Ingle, “Smart Traffic: Integrating Machine Learning, and YOLO for Adaptive Traffic Management System”,International Journal of Intelligent Systems and Applications in Engineering,VOL. 12 NO. 12S (2024). [3] Dakshayani Ijeri, Prasanna Maidargi, Raju Sunagar, “Traffic Control System Using Image Proccesing”. [4] Khaled Zaatouri, Tahar Ezzedine, “A Self-Adaptive Traffic Light Control System Based on YOLO”, International Conferences on Internet of Things, Embedded Systems and Communications, December 2018. [5] C S Asha,A V Narasimhadhan, “Vehicle Counting for Traffic Management System using YOLO and Correlation Filter”,IEEE, International Conference on Electronics, Computing and Communication Technologies (CONECCT),March 2018. [6] K. Vidhya, A. Bazila Banu, “Density Based Traffic Signal System”, International Journal of Innovative Research in Science,Engineering and Technology, Volume 3, Special Issue 3, March 2014. [7] Roxanne Hawi, George Okeyo, Michael Kimwele, “Techniques for Smart Traffic Control: An In-depth Review”, International Journal of Computer Applications Technology and Research, Volume 4– Issue 7, 566 - 573, 2015, ISSN: 2319–8656.

Copyright

Copyright © 2025 Dr. Prakash N. Kalavadekar, Avanti Joshi, Vaishnavi Mohadkar, Pranjal Jagtap, Savari Jawale. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.