The invention discloses a DQN and CNN-based expressway automatic driving exclusive lane vehicle confluence method, and belongs to the technical field of automatic driving. The method mainly comprises the following steps: building a reinforcement learning model according to a one-way multi-lane scene of an expressway, comprising the steps that a state space obtained based on detection area road gridding processing is established, an action space is set based on whether opening of a lane changing key area and duration time, and a reward function is constructed to increase the number of lane changing vehicles in average green light time and reduce the opening cost of lane changing signals; and the CNN-based intelligent agent adopts a DQN algorithm to solve an afflux strategy, a DQN-CNN-based deep reinforcement learning model is trained, and an intelligent afflux model of the vehicles on the automatic driving exclusive lane of the expressway is obtained. The deep reinforcement learning algorithm is utilized to improve the safety and efficiency when the automatic driving vehicle merges into the special lane in the highway scene, and the complex and changeable traffic problem in the highway system can be better solved.

    本发明公开了一种基于DQN和CNN的高速公路自动驾驶专用道车辆汇入方法,属于自动驾驶技术领域。本发明方法主要包括:依据高速公路单向多车道场景搭建强化学习模型,包括建立基于检测区道路网格化处理得到的状态空间、基于换道关键区的开放是否和持续时间设置动作空间、构建奖励函数以增加平均绿灯时间内的换道车辆数和降低换道信号开放成本;基于CNN的智能体采用DQN算法进行汇入策略求解,训练基于DQN‑CNN的深度强化学习模型,获得高速公路自动驾驶专用道车辆智能汇入模型。本发明利用深度强化学习算法提高了高速公路场景下自动驾驶车辆汇入专用道时的安全性和高效性,可以更好地解决高速公路系统中复杂多变的交通问题。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    DQN and CNN-based expressway automatic driving exclusive lane vehicle afflux method


    Weitere Titelangaben:

    一种基于DQN和CNN的高速公路自动驾驶专用道车辆汇入方法


    Beteiligte:
    ZHANG JIAN (Autor:in) / XIONG ZHUANG (Autor:in) / JIANG HAN (Autor:in) / LI YAN (Autor:in) / ZHANG YUJIE (Autor:in) / SHEN AIZHONG (Autor:in) / YAO JINQIANG (Autor:in)

    Erscheinungsdatum :

    2024-05-03


    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



    Vehicle ramp afflux-in and afflux-out cooperative control method

    SHAN PING | Europäisches Patentamt | 2022

    Freier Zugriff

    Vehicle afflux decision-making method considering equality of afflux motivation

    LIU JINQIANG / ZHAO WANZHONG / WANG CHUNYAN et al. | Europäisches Patentamt | 2023

    Freier Zugriff

    Automatic driving vehicle high-speed ramp intelligent afflux method based on reinforcement learning

    HU XIAOSONG / XIAO WENXUAN / YANG YUYOU et al. | Europäisches Patentamt | 2023

    Freier Zugriff

    Method and device for determining forced lane changing afflux point of intelligent vehicle

    DONG CHANGYIN / WANG HAO / QU YING et al. | Europäisches Patentamt | 2021

    Freier Zugriff

    Ramp afflux-in and afflux-out early warning system

    YANG LEI / HONG TAO / WANG HENGDA et al. | Europäisches Patentamt | 2021

    Freier Zugriff