The invention discloses an intersection vehicle trajectory optimization method and system based on off-line reinforcement learning, and the method comprises the following specific steps: building a communication scene of intersection signal lamps and network connection vehicles in an intelligent network connection environment, setting a communication distance threshold value of the vehicles and the signal lamps in the scene, and when the vehicles drive to a control region of the signal lamps, carrying out the control of the signal lamps; and the two parties can perform information interaction. The method comprises the following steps: collecting running tracks and signal phase timing data of network-connected vehicles at a signalized intersection, abstracting a running process of the vehicles at the intersection into a Markov decision process, taking vehicle travel time, vehicle energy consumption, collision time and the like as parameters of a Markov decision reward function, and constructing an applicable data set; and training the vehicle intelligent agent through a reinforcement learning method in an off-line state to obtain an applicable intersection passing strategy. The method can be applied to vehicle control at the intersection, the vehicle runs according to the optimal track by controlling the acceleration of the vehicle, and energy conservation, emission reduction and safety performance improvement of the vehicle are achieved.

    本发明公开了一种基于离线强化学习的交叉口车辆轨迹优化方法、系统,具体步骤如下:搭建智能网联环境下交叉口信号灯与网联车辆的通信场景,设定场景内车辆与信号灯的通信距离阈值,当车辆行驶至信号灯的控制区域时,双方可进行信息交互。收集信号交叉口网联车辆的运行轨迹和信号相位配时数据,将车辆在交叉口的行驶过程抽象为马尔科夫决策过程,将车辆旅行时间、车辆能耗、碰撞时间等作为马尔科夫决策奖励函数的参数,构建适用的数据集,在离线状态下通过强化学习方法训练车辆智能体得到适用的交叉口通行策略。本发明可应用于交叉口的车辆控制,通过控制车辆加速度使车辆按照最优轨迹行驶,实现车辆的节能减排和安全性能提升。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Intersection vehicle trajectory optimization method and system based on offline reinforcement learning


    Weitere Titelangaben:

    一种基于离线强化学习的交叉口车辆轨迹优化方法、系统


    Beteiligte:
    ZHANG JIAN (Autor:in) / SHI XIAOYU (Autor:in) / JIANG XIA (Autor:in) / ZHANG HAIYAN (Autor:in) / WANG BO (Autor:in) / QIAN PINZHENG (Autor:in)

    Erscheinungsdatum :

    2023-08-11


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

    IPC:    G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen