A Method of Reinforcement Learning Based Automatic Traffic Signal Control

To improve performance of traffic signal control system in urban area, a novel method is proposed in this paper. The roads, vehicles and the traffic control systems are all modeled as intelligent agents. Wireless communication network provides the possibility of the cooperation of vehicles and roads. Based on all the information from vehicles and roads, a traffic control policy can be planned online according to the updated situation on the roads. The optimum intersection signals can be learned automatically on line based on reinforcement learning. An intersection signal control system is studied as an example of the method with a Q-learning based algorithm. The simulation results show that the proposed intersection signal control can improve traffic efficiency by about 30% over the traditional method.