TRAFFIC LIGHT MONITORING AND CONTROL USING CNN

TRAFFIC LIGHT MONITORING AND CONTROL USING CNN

Sugha S1, Akzin Jayas A J2, Kishore R3, Chandru K S4

 Student(1), Student(2), Student(3), Faculty(4)

sugha.cb20@bitsathy.ac.in 1, akzinjayas.cb20@bitsathy.ac.in 2, kishore.cb20@bitsathy.ac.in 3,

chandruks@bitsathy.ac.in 4

Bannari Amman Institute of Technology, Sathyamangalam.

Abstract:
Urban traffic congestion plagues cities worldwide, resulting in wasted fuel, increased travel times, and environmental pollution. Traditional traffic light systems with fixed timings are ineffective at adapting to real-time traffic fluctuations. This project proposes a novel Intelligent Traffic Light Control (ITLC) system that leverages the power of computer vision to optimize traffic flow at intersections. The core of the ITLC system hinges on a two-pronged approach utilizing Convolutional Neural Networks (CNNs) and Temporal Convolutional Networks (TCNs). First, CNNs, trained on extensive datasets of traffic camera images, will be adept at accurately detecting and counting vehicles in each lane. Subsequently, TCNs will analyze video feeds to capture the temporal dynamics of traffic flow, understanding the evolving patterns of vehicle movement over time. By fusing information from both static images and video analysis, the ITLC system gains a comprehensive understanding of the real-time traffic scenario at an intersection. This data is then fed into a Q-Learning algorithm, which dynamically adjusts traffic light timings based on the current number of vehicles and historical traffic patterns. This intelligent approach prioritizes congested lanes, optimizes overall traffic flow, and minimizes unnecessary delays. The ITLC system strives to significantly reduce wait times at intersections, leading to a smoother driving experience and reduced fuel consumption. It also promises to improve traffic flow by dynamically adapting to changing conditions, maximizing the capacity of each intersection. The success of this project has the potential to revolutionize urban traffic management, paving the way for a more efficient and environmentally friendly transportation network in cities around the globe.

Keywords: Convolutional Neural Networks (CNNs), Temporal Convolutional Networks (TCNs), Q-Learning, Traffic Flow Optimization.