Deep Deterministic Policy Gradient for Traffic Signal Control of Single Intersection

This paper aims to apply deep reinforcement learning algorithms to signal control at single intersection to relieve traffic congestion. A deep deterministic policy gradient (DDPG) based method in which continuous actions and variable phases sequence and cycle time adopted is proposed to adjust the sequence and duration of signal phases to minimize average vehicle delay time. The performance of the proposed method is compared with deep Q network (DQN) and fix-time control. Simulation results indicate that DDPG algorithm is superior to DQN algorithm and fix-time control method in dynamic traffic environment.