The invention discloses a mountain road automatic driving control method and system, and relates to the field of automatic driving, the method is applied to the mountain road automatic driving control system, and the method comprises the steps: building a mountain road simulation environment through an automatic driving simulator TORCS, and traversing the mountain road simulation environment to design an automatic driving task set; converting the automatic driving task set into a Markov decision tuple; the method comprises the following steps: establishing a TAD3 algorithm model by utilizing a PyTorch framework based on a python language; and according to the Markov decision tuple, training the TAD3 algorithm model based on an actor-commentator algorithm, and generating a mountain road automatic driving control agent. The technical problem that in the prior art, because a traditional reinforcement learning mode cannot rapidly obtain an agent suitable for automatic driving control of a mountainous area scene, an agent which can be rapidly trained and determined and is high in adaptability to mountainous area roads does not exist is solved.
本申请公开了一种山路自动驾驶的控制方法及系统,涉及自动驾驶领域,方法应用于一山路自动驾驶的控制系统,包括:通过自动驾驶仿真器TORCS建立山路模拟环境,遍历山路模拟环境设计自动驾驶任务集合;将自动驾驶任务集合转换为马尔可夫决策元组;基于python语言,利用PyTorch框架,建立TAD3算法模型;根据马尔可夫决策元组,基于演员‑评论家算法,对TAD3算法模型进行训练,生成山路自动驾驶控制智能体。解决了现有技术中由于传统的强化学习方式无法快速得到适用于山区场景的自动化驾驶控制的智能体,导致存在缺乏可快速训练确定,且对山区道路适应性较强的智能体的技术问题。
Mountain road automatic driving control method and system
一种山路自动驾驶的控制方法及系统
2022-11-22
Patent
Electronic Resource
Chinese
Control method and system for mountain road driving of electric vehicle
European Patent Office | 2024
|Suspension control method and device for driving on mountain road
European Patent Office | 2024
|Digital reconstruction automatic driving centralized road control system
European Patent Office | 2024
|Bidirectional driving early warning system for narrow road of mountain
European Patent Office | 2023
|