基于Q学习的协同多路径规划研究

Path planning for vehicles is an effective means to solve urban traffic congestion.The traditional path planning algorithms usually give only the optimal path. Once the optimal path is invalid, no alternative paths exist for vehicles. In order to improve the stability of path planning, multiple paths planning technology is introduced to ensure that at least one optional path exists for vehicles in any case.This paper proposes a Q-learning based cooperative algorithm with FNN to obtain optimal multiple paths set for Origin-Destination (OD) pairs. First, a Q-learning algorithm is applied to represent the road model. Second, the return value of Q-learning is skillfully used to find multiple paths. Reliable constraints of multiple paths set are introduced to ensure at least one available path exists in any case. Furthermore, the cooperative mechanism is introduced to coordinate path planning among agents to balance the future traffic load. Simulation results validate its efficiency and reliability.