The invention relates to the technical field of vehicle driving control, in particular to a vehicle path tracking coordination control method and system based on danger level judgment and reinforcement learning. The method comprises the following steps: acquiring an environment state quantity S0 of a target vehicle at the current moment and a danger level of the target vehicle at the current moment; calculating the next action a of the target vehicle according to the S0 by using a reinforcement learning algorithm; selecting a corresponding danger penalty term Rd according to the current danger level of the target vehicle, and adding the danger penalty term Rd into a reward of a reinforcement learning algorithm to form a reward function R for calculation; and the optimal front wheel steering angle increment # imgabs0 # optimal yawing moment increment delta M * at the next moment is calculated according to a, and output control is carried out on the optimal front wheel steering angle increment # imgabs0 # optimal yawing moment increment delta M *. According to the method, steering and differential braking are coordinated based on reinforcement learning, a danger level penalty term is added in reinforcement learning, control over the intervention opportunity and the intervention intensity of the yawing moment is achieved, and the problem of insufficient steering can be effectively solved.
本发明涉及车辆驾驶控制技术领域,具体涉及基于危险等级判别和强化学习的车辆路径跟踪协调控制方法、系统。本发明的方法包括:获取目标车辆当前时刻环境状态量S0、及目标车辆当前时刻所处的危险等级;使用强化学习算法依据S0计算出目标车辆下一步动作a;根据目标车辆当前所处的危险等级选择对应的危险惩罚项Rd,并加入到强化学习算法的奖励中,形成用于计算的奖励函数R;依据a计算下一时刻的最优前轮转角增量#imgabs0#最优横摆力矩增量ΔM*,并输出控制。本发明基于强化学习来协调转向和差动制动,并在强化学习中加入了危险等级惩罚项,对横摆力矩实现介入时机、介入强度的控制,可以有效减轻转向不足的问题。
Vehicle path tracking coordination control method and system based on danger level judgment and reinforcement learning
基于危险等级判别和强化学习的车辆路径跟踪协调控制方法、系统
09.08.2024
Patent
Elektronische Ressource
Chinesisch
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 |
Europäisches Patentamt | 2020
|Path tracking stability control method based on reinforcement learning
Europäisches Patentamt | 2024
|Europäisches Patentamt | 2024
|Vehicle Traveling Data Recorder Design Based on Judgment of Danger
Trans Tech Publications | 2014
|Deep reinforcement learning based path tracking controller for autonomous vehicle
SAGE Publications | 2021
|