Executive Summary : | The increasing population is leading to urban traffic congestion on roads, making the static timer traffic control system a significant challenge. Traffic police can stop, slow, and safely direct traffic, but human limitations such as low vision make it difficult to control traffic at night. Heavy traffic from all directions is also a challenge for traffic police. The main challenge in traffic control systems is to efficiently accommodate traffic and reduce waiting times at junctions. Static timer control signals direct vehicles to cross junctions but fail to reduce delays in quick movement of traffic. Traffic at junctions is unpredictable, and fixed green time for each lane doesn't reduce queue length for high vehicle density lanes. Zero density lanes also get the same amount of green time, creating unnecessary waiting time for other lanes. Various solutions have been proposed to solve congestion problems on traffic junctions, including using an RL-based algorithm in Adaptive Traffic Signal Control (ATSC). Technologies like Deep Neural Network (DNN) and Simulation of Urban Mobility are used to improve traffic control systems. DNN works as the best functional approximates, estimating optimal policies through continuous learning. The existing static time traffic control system fails to effectively manage traffic at junctions due to its inability to recognize lanes with high vehicular density, leading to wastage of green signal time and high congestion on other lanes. To improve traffic management, the ATSC using RL algorithm is adapted, achieving objectives by exploring the environment and updating the policy. |