Q-learning Based Traffic Signal Control Method for an Isolated Intersection
Bao-Lin Ye,Peng Wu,Weimin Wu,Lingxi Li,Yaodong Zhu,Bin Chen
DOI: https://doi.org/10.1109/cac57257.2022.10054839
2022-01-01
Abstract:Traffic signal control strategy plays an important role in determining the traffic efficiency of intersections. In order to increase the capacity of intersections, we propose an improved Q-learning based method for traffic signal control. Firstly, we define a novel reward function consisting of two parts to dynamically adjust the timing plan of an isolated intersection by utilizing the real-time traffic situation of the intersection. In the reward function, the first part is employed to reflect the change of traffic state of each lane, while the second part is essentially employed to adjust the signal timing of each phase. In addition, a vehicle position information matrix is defined to describe the traffic state of the intersection at each time step. Finally, the simulation and experimental results based on VISSIM demonstrate the validity of the proposed method.