【作者】 张发旺； 谢根金； 于光远； 万航； 聂士达； 刘辉；

【Author】 Zhang Fawang;Xie Genjin;Yu Guangyuan;Wan Hang;Nie Shida;Liu Hui;School of Mechanical Engineering,Beijing Institute of Technology;Jingwei Hirain(Tianjin)Research&Development Co.,Ltd;School of Mechanical Engineering,University of Science and Technology Beijing;

【机构】 北京理工大大学机械与车辆学院； 经纬恒润(天津)研究开发有限公司； 北京科技大学机械工程学院；

【摘要】 车辆底盘域高实时控制是实现辅助驾驶、高级别自动驾驶的关键技术。现有的以滚动时域优化为典型代表的控制类策略求解方法存在计算量大、求解缓慢等问题,无法满足车辆控制的实时性需求。针对该问题,本文通过牛顿-欧拉法建立了车辆底盘14自由度动力学模型,针对底盘域控制问题中的车道保持控制和自适应巡航控制两个典型工况,构建对应的最优控制问题形式,利用强化学习方法求解得到可满足计算实时性要求的最优控制策略,并通过硬件在环仿真验证了最优策略的求解实时性。本研究为底盘域高实时控制策略的进一步开发和应用奠定了基础。更多还原

【Abstract】 High real-time control in vehicle chassis domain is a key technology for achieving assisted driving and high-level autonomous driving.The existing control strategy solving methods,typically represented by receding horizon optimization,have problems such as large computational complexity and slow solution,which cannot meet the real-time requirements of vehicle control.To solve this problem,this paper establishes a high fidelity vehicle chassis dynamics model,designs an optimal control problem for high real-time control in the chassis domain,and uses reinforcement learning methods to solve the optimal control strategy that can meet the requirements of real-time computing.Hardware in loop simulation results verify the real-time solution of the optimal strategy.This study lays the foundation for the further development and application of high real-time control strategies in the chassis domain.更多还原