With the continuous advancement of urbanization, traffic congestion becomes a major challenge for social development. Intelligent driving technology can ease the traffic pressure at intersections, which is the trend of future traffic development. Cooperative Adaptive Cruise Control (CACC) is a specific application of intelligent driving technology and a necessary step toward autonomous driving. This paper proposes a signal control method for CACC homogeneous traffic flow at single intersections based on deep reinforcement learning, which considers the operation characteristics of CACC vehicles at intersections and uses a grid-based state space input to improve the model prediction accuracy. On this basis, this paper proposes a variable phase duration strategy to enhance the phase control flexibility and a platoon clearance time to reduce the vehicle travel delay. Simulation experiments show that the signal control method based on deep reinforcement learning has a significant advantage over traditional signal control methods in the CACC homogeneous traffic flow environment and is more suitable for intersections with high traffic volume. The research results of this paper not only contribute to enriching the related research field but also offer a valuable reference for improving traffic safety and efficiency at intersections.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Optimizing Intersection Signal Via Reinforcement Learning for Cooperative Adaptive Cruise Control


    Contributors:
    Yang, Shu (author) / Yuan, Rui (author) / Wang, Miao (author) / Sun, Liang (author) / Wu, Jingyuan (author)


    Publication date :

    2023-10-28


    Size :

    1464828 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English






    Cooperative Adaptive Cruise Control Based on Intent Sharing Messages and Reinforcement Learning

    Kanipakam, Lahiri / Sakr, Ahmed Hamdi / Avedisov, Sergei S. et al. | IEEE | 2024