The invention discloses an automatic driving decision uncertainty estimation method based on reinforcement learning, and relates to the technical field of automatic driving, and the method comprises the following steps: obtaining intelligent agent related data, and constructing an MDP model; constructing a neural network model set; constructing a neural network model set, and inputting the state s into a neural network model to obtain an action-value function Q; the intelligent agent selects the optimal action according to the Q value, and uncertainty estimation or autonomous driving decision making is achieved; in the previous research, the intelligent agent trained through reinforcement learning can make a reasonable decision under the condition of approaching training data, so that the effectiveness of the reinforcement learning technology in decision making is verified to a certain extent; according to the method, the uncertainty of the recommended action can be effectively estimated, and the action with the small risk can be selected according to the uncertainty estimation in the face of an unknown condition; due to the characteristic, the method has remarkable advantages in processing tasks which need to face complicated and uncertain environments, such as automatic driving and the like.

    本发明公开一种基于强化学习的自动驾驶决策不确定性估计方法,涉及自动驾驶技术领域,包括如下步骤:获取智能体相关数据,构建MDP模型;构建神经网络模型集合;构建神经网络模型集合,将状态s输入神经网络模型以获得行动‑价值函数Q;智能体根据Q值选择最优行动,实现不确定性估计或自主驾驶决策;在以往的研究中,通过强化学习训练出的智能体能够在接近训练数据的情况下做出合理的决策,这在一定程度上验证了强化学习技术在决策制定上的有效性;本发明方法不仅能够有效地估计推荐行动的不确定性,而且能够根据这种不确定性估计在面临未知情况时选择风险较小的行动;这一特性使其在处理自动驾驶等需要面临复杂、不确定环境的任务时具有显著优势。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Automatic driving decision uncertainty estimation method based on reinforcement learning


    Weitere Titelangaben:

    一种基于强化学习的自动驾驶决策不确定性估计方法


    Beteiligte:
    ZOU QIANYING (Autor:in) / LIAO YUAN (Autor:in)

    Erscheinungsdatum :

    2023-10-10


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

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




    Automatic driving decision-making method based on deep reinforcement learning

    LIU CHENGQI / LIU SHAOWEIHUA / ZHANG YUJIE et al. | Europäisches Patentamt | 2024

    Freier Zugriff

    Automatic driving decision-making method based on rule-assisted reinforcement learning

    ZHENG KAI / SU HAN / ZENG XIMU | Europäisches Patentamt | 2023

    Freier Zugriff

    Automatic driving vehicle microscopic decision-making method based on reinforcement learning

    ZHENG KAN / LIU JIE / ZHAO LONG | Europäisches Patentamt | 2020

    Freier Zugriff

    Automatic driving behavior decision-making method based on deep reinforcement learning

    YANG MINGZHU / LIU XIANGWEI / LI ZHUOLUO | Europäisches Patentamt | 2020

    Freier Zugriff