The invention belongs to the technical field of automobiles, and particularly relates to an intelligent automobile roundabout driving decision-making method based on behavior risk reinforcement learning. The method comprises the following steps: firstly, analyzing and establishing an automobile transverse and longitudinal coupling action space, a multi-scale state space containing a driving task-surrounding environment-automobile information, and a multi-target reward function considering safety-task-driving behaviors; secondly, classifying and evaluating agent behavior risks by applying a self-learning behavior risk classifier of a multi-layer perceptron, and designing a strategy optimization target based on the behavior risks on the basis of a behavior risk classification result to optimize a reinforcement learning driving decision algorithm so as to solve the influence of insufficient design of a reward function on driving strategy optimization and improve the driving strategy optimization accuracy. And the driving decision is guided to converge towards a safer direction.

    本发明属于汽车技术领域,具体的说是一种基于行为风险强化学习的智能汽车环岛驾驶决策方法。首先,分析并建立汽车横纵向耦合动作空间、包含驾驶任务‑周围环境‑自车信息的多尺度状态空间、考虑安全‑任务‑驾驶行为的多目标奖励函数。其次,通过应用多层感知机的自学习行为风险分类器,对智能体行为风险进行分类评定,并基于行为风险分类结果设计基于行为风险的策略优化目标,来优化强化学习驾驶决策算法,以解决奖励函数设计不足对于驾驶策略优化的影响,并引导驾驶决策朝向更安全方向收敛。


    Access

    Download


    Export, share and cite



    Title :

    Intelligent automobile roundabout driving decision-making method based on behavior risk reinforcement learning


    Additional title:

    一种基于行为风险强化学习的智能汽车环岛驾驶决策方法


    Contributors:
    WU JIAN (author) / SHI YUKANG (author)

    Publication date :

    2024-11-22


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    B60W CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION , Gemeinsame Steuerung oder Regelung von Fahrzeug-Unteraggregaten verschiedenen Typs oder verschiedener Funktion / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen




    Roundabout multi-vehicle collaborative decision-making method based on safety reinforcement learning

    XIE XIANYI / HU TAO / JIN LISHENG et al. | European Patent Office | 2024

    Free access

    Intelligent Roundabout Insertion using Deep Reinforcement Learning

    Capasso, Alessandro Paolo / Bacchiani, Giulio / Molinari, Daniele | ArXiv | 2020

    Free access

    Intelligent automobile decision-making method based on driving intention and deep reinforcement learning

    PEI XIAOFEI / LU SONGXIN / YANG BO | European Patent Office | 2024

    Free access

    Parameter Sharing Reinforcement Learning for Modeling Multi-Agent Driving Behavior in Roundabout Scenarios

    Konstantinidis, Fabian / Hofmann, Ulrich / Sackmann, Moritz et al. | IEEE | 2021