TK1110 : Intelligent management of urban intersections using reinforcement learning and vehicle counting data
Thesis > Central Library of Shahrood University > Electrical Engineering > MSc > 2026
Authors:
[Author], Seyed Masoud Mirrezaei[Supervisor]
Abstarct: The continuous growth of urban traffic demand and the limited capacity of transportation infrastructures have made intelligent intersection management one of the major challenges in Intelligent Transportation Systems (ITS). Conventional traffic signal control approaches are mainly baxsed on fixed-time scheduling or predefined models and often fail to adapt effectively to dynamic traffic conditions. This thesis proposes an intelligent real-time urban intersection management frxamework baxsed on Deep Reinforcement Learning (DRL) and vehicle counting data. In the proposed approach, real-time traffic measurements obtained from vehicle counting sensors are utilized as the environmental state, while a learning agent employs the Deep Q-Network (DQN) algorithm to dynamically determine the optimal traffic signal control policy. To evaluate the effectiveness of the proposed method, a four-leg urban intersection is modeled in the Simulation of Urban Mobility (SUMO) environment under different traffic scenarios, including low, medium, and high traffic demand conditions. The state space is defined using vehicle counts, queue lengths, and the current signal phase, whereas the reward function is designed to minimize vehicle waiting time, reduce queue lengths, and maximize traffic throughput. Simulation results demonstrate that the proposed frxamework significantly outperforms conventional fixed-time and adaptive signal control methods in terms of average delay reduction, queue length minimization, and overall intersection efficiency. The findings indicate that deep reinforcement learning driven by real-time traffic data can provide an effective solution for intelligent and online control of urban intersections and can contribute to the development of next-generation smart transportation systems.
Keywords:
#Keywords: Intelligent Traffic Management #Deep Reinforcement Learning #Deep Q-Network (DQN) #Traffic Signal Control #Urban Intersection #Vehicle Counting Data #Intelligent Transportation Systems Keeping place: Central Library of Shahrood University
Visitor: