Q-learning based intelligent multi-objective particle swarm optimization of light control for traffic urban congestion management

Traffic signal operations play an important role in the effective functioning of the urban area. However, due to the increasing number of vehicles and the high dynamic of the traffic network, conventional traffic signal timing methods does not result in an efficient control. One alternative is to let traffic signal controllers learn how to adjust the lights based on the traffic situation. In this paper, we propose a novel multi-objective traffic light control system that is based on an Intelligent Multi-Objective Particle Swarm Optimization (MOPSO) method. We take the average junction waiting time and the flow rate of vehicles on the congested road as two objectives. In the proposed method, we granted the ability of selecting appropriate MOPSO parameters to each agent of the swarm via a novel multi-objective Q-Learning approach. The simulation results demonstrate the efficiency of the proposed system.