The invention discloses an automatic driving lane selection decision-making method and system based on inverse reinforcement learning. The method comprises the following steps: constructing an expert strategy by using a planning algorithm, and performing simulation in a highway-env environment to obtain an expert data set; secondly, reconstructing a local reward function on the basis of an action device-evaluator network architecture of reinforcement learning: measuring the uncertainty of an intelligent agent to a state by utilizing an extended hybrid evaluation network pool in a reinforcement learning process so as to select a curriculation sub-target; and based on the sub-target reconstruction local matching reward function, guiding the agent to act according to the expert trajectory. And finally, updating a local internal reward generator by utilizing a meta-imitation learning method in a reinforcement learning process based on a meta-imitation target, guiding the intelligent agent to explore near expert data, and reasoning and modeling expert intention from the perspective of a reward function, so as to further guide the intelligent agent to approach an expert strategy.
基于逆强化学习的自动驾驶选道决策方法和系统,其方法包括:利用规划算法构造专家策略在highway‑env环境中进行仿真得到专家数据集。然后,在强化学习的行动器‑评判器网络架构基础上重构局部奖励函数:在强化学习过程中利用扩展混合评判网络池衡量智能体对于状态的不确定度进而选择课程化子目标;基于子目标重构局部匹配奖励函数引导智能体按照专家轨迹行动。最后利用基于元模仿目标,在强化学习过程中间利用元模仿学习的方法更新局部内在奖励生成器,引导智能体在专家数据附近探索,从奖励函数角度推理建模专家意图,从而进一步引导智能体靠近专家策略。
Automatic driving lane selection decision-making method and system based on inverse reinforcement learning
基于逆强化学习的自动驾驶选道决策方法和系统
2023-10-17
Patent
Electronic Resource
Chinese
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 |
Automatic driving lane keeping decision-making method based on deep reinforcement learning
European Patent Office | 2025
|Hybrid automatic driving lane changing decision-making method based on deep reinforcement learning
European Patent Office | 2024
|Lane-Change Decision of Automatic Driving Based on Reinforcement Learning Framework
Transportation Research Record | 2024
|European Patent Office | 2022
|Reinforcement learning multi-lane driving decision-making method for dynamic traffic environment
European Patent Office | 2024
|