Traffic congestion is a serious problem that needs to be solved in the development of intelligent transportation, and global path planning is an effective way, especially with the collaborative optimization on various traffic participants. In this paper, we formulate the vehicle global path planning as a Markov decision problem, and propose an $\mathbf{A}^{\ast}$ -assisted deep reinforcement learning algorithm to achieve path optimization in complex environment by maximizing the total rewards. Compared with the existing ones in the literature, the proposed method can not only filter and extract unobvious features from multidimensional observations, but also make dynamic decisions in complex environment. To verify the effectiveness of our algorithm, we design experiments on the Simulation of Urban MObility (SUMO) platform. It is shown that the proposed algorithm converges faster and can deal with the changing and complex environment more effectively, which provides a practical solution to the problem of urban traffic congestion.
Vehicle-Road Collaborative Path Planning Based on $\mathbf{A}^{\ast}$ -assisted Deep Reinforcement Learning
08.10.2022
2497195 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Europäisches Patentamt | 2024
|ArXiv | 2023
|