The invention discloses an intelligent automobile decision-making method based on driving intention and deep reinforcement learning, and the method comprises the steps: building an intention reasoning model through a neural network, and making a data set; the method comprises the following steps: initializing intent inference neural network parameters, and setting neural network training hyper-parameters to obtain a convergent intent inference model; the method comprises the following steps: building reinforcement learning decision models in different driving scenes by using a neural network for an automatic driving vehicle based on a reinforcement learning RainbowDQN algorithm; acquiring current vehicle states and historical perception information of the intelligent vehicle and surrounding vehicles, reasoning driving intentions of surrounding drivers by using an intention reasoning model, and tagging surrounding vehicle types and steering signals as auxiliary driving intention discrimination; and calling corresponding convergent reinforcement learning decision models for different driving scenes. According to the method, the driving intention of the driver is obtained through intention reasoning so as to judge the uncertainty of behaviors of the driver, the uncertainty is considered in decision making, and the learning efficiency of reinforcement learning and the high efficiency of decision making are improved.
本发明公开了基于驾驶意图和深度强化学习的智能汽车决策方法包括,利用神经网络搭建意图推理模型,制作数据集;通过初始化意图推理神经网络参数,对神经网络训练超参数进行设定,获得收敛的意图推理模型;面向自动驾驶汽车基于强化学习RainbowDQN算法,利用神经网络搭建不同驾驶场景下的强化学习决策模型;获取智能汽车和周围车辆的当前车辆状态与历史感知信息,利用意图推理模型推理周围驾驶员的驾驶意图,将周围车辆类型与转向信号进行标签化作为辅助驾驶意图辨别;对于不同的驾驶场景调用对应收敛的强化学习决策模型。本方法利用意图推理获取驾驶员驾驶意图,以判断其行为的不确定性,并在决策中加以考虑,提高强化学习的学习效率与决策的高效性。
Intelligent automobile decision-making method based on driving intention and deep reinforcement learning
基于驾驶意图和深度强化学习的智能汽车决策方法
2024-10-22
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 / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen |
Deep Reinforcement Learning with Driving Intention for On-Ramp Decision-Making
Springer Verlag | 2025
|European Patent Office | 2024
|Automatic driving decision-making method based on deep reinforcement learning
European Patent Office | 2024
|Intelligent driving decision-making system and method based on reinforcement learning
European Patent Office | 2025
|Automatic driving behavior decision-making method based on deep reinforcement learning
European Patent Office | 2020
|