The invention discloses an ethical-driven multi-mode decision-making method based on deep reinforcement learning, and the method comprises the steps: obtaining a vehicle shooting image, and obtaining the morphological characteristics and dynamic characteristics of a surrounding environment based on the vehicle shooting image and a perception model; constructing a multi-modal neural network, introducing an ethical coefficient fed back by human ethics into the multi-modal neural network for training, and then calculating the morphological characteristics and the dynamic characteristics of the surrounding environment to obtain a Q value of a corresponding action; and calculating the Q value of the corresponding action based on a reinforcement learning algorithm to obtain the corresponding action, and making a decision on the vehicle based on the corresponding action. According to the method, the object morphological characteristics observed in the environment and the dynamic characteristics of the relative state of the vehicle obtained through the vehicle-mounted sensor are used as model input, and the ethical-driven multi-mode neural network is provided based on the model input to uniformly extract the two morphological characteristics, so that the accuracy and robustness of the overall state information can be improved.

    本发明公开了一种基于深度强化学习的伦理驱动多模态决策方法,包括:获取车辆拍摄图像,基于所述车辆拍摄图像和感知模型获取周围环境形态特征和动态特征;构建多模态神经网络,将人类伦理反馈的伦理系数引入至所述多模态神经网络中进行训练后对所述周围环境形态特征和动态特征进行计算,获得相应动作的Q值;基于强化学习算法将所述相应动作的Q值进行计算,获得对应动作,基于所述对应动作对车辆进行决策。本发明采用在环境中观测到的物体形态特征与通过车载传感器获取车辆相对状态的动态特征作为模型输入,并基于此提出了伦理驱动的多模态神经网络统一提取两种形态特征,可以提升整体状态信息的准确性和鲁棒性。


    Access

    Download


    Export, share and cite



    Title :

    Deep reinforcement learning-based ethical-driven multi-modal decision-making method


    Additional title:

    一种基于深度强化学习的伦理驱动多模态决策方法


    Contributors:
    LI XUEYUAN (author) / GAO XIN (author) / LUAN TIAN (author) / MENG XIAOQIANG (author) / LIU QI (author)

    Publication date :

    2023-10-20


    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 / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G06V




    Cascade deep reinforcement learning security decision-making method based on multi-modal space-time representation

    YANG YUXIANG / GE FENGLONG / ZHAO JUFENG et al. | European Patent Office | 2024

    Free access

    Intersection decision-making method based on multi-agent deep reinforcement learning

    DU YU / JIANG ANNI / ZHAO SHIXIN et al. | European Patent Office | 2024

    Free access

    Lane keeping decision-making method based on deep reinforcement learning

    WANG PENGWEI / ZHOU HENGHENG / GAO SONG et al. | European Patent Office | 2023

    Free access

    Automatic driving decision-making method based on deep reinforcement learning

    LIU CHENGQI / LIU SHAOWEIHUA / ZHANG YUJIE et al. | European Patent Office | 2024

    Free access