The invention is applicable to the technical field of intelligent traffic systems, and provides an automatic road intersection management method based on projection constraint strategy optimization, which introduces a risk network, divides safety levels during strategy updating, maximizes a reward value on the premise of meeting safety, and improves the efficiency of road intersection management. The problem that a reinforcement learning algorithm violates safety when pursuing passing efficiency and comfort is solved. In addition, the strategy neural network uses the LSTM neural network to capture the relationship between the vehicles and centrally regulate and control all the vehicles in the complex intersection so as to obtain a strategy with more excellent performance. Through experimental tests, the performance of the method is compared with the performance of PPO and PPO-SC based on non-risk perception, the performance of a vehicle intersection coordination scheme algorithm based on MPC and the performance of an MICA algorithm based on MIP, and experiments show that the method has more excellent performance.
本发明适用于智能交通系统技术领域,提供了一种基于投影式约束策略优化的自动道路交叉口管理方法,引入风险网络,在更新策略时划分安全等级,在满足安全性的前提下最大化奖励值,解决了强化学习算法在追求通行效率与舒适性会违背安全性的问题。除此之外,策略神经网络应用LSTM神经网络捕获每辆车之间的关系并且集中调控复杂路口中的全部车辆以获得性能更加优良的策略。通过实验测试,并将与基于非风险感知的PPO和PPO‑SC、基于MPC的车辆交叉口协调方案算法以及基于MIP的MICA算法的性能进行对比,实验表明本方法具有更加优秀的表现。
Automatic road intersection management method based on projection type constraint strategy optimization
一种基于投影式约束策略优化的自动道路交叉口管理方法
2023-10-24
Patent
Electronic Resource
Chinese
Vehicle adaptive cruise control method based on projection constraint strategy optimization
European Patent Office | 2023
|European Patent Office | 2021
|Automatic driving method and system based on experience playback constraint strategy optimization
European Patent Office | 2024
|