无人车辆横向跟踪控制研究
Research on Lateral Tracking Control of Autonomous Vehicles
李伯雄 1王立勇 1孙鹏 1季文龙2
作者信息
- 1. 北京信息科技大学现代测控技术教育部重点实验室,北京 100192
- 2. 中国人民解放军63966部队,北京 100072
- 折叠
摘要
使用传统模型预测控制对车辆轨迹进行跟踪时,模型中的路面附着系数往往为特定工况下的经验数值.当车辆在未知路面行驶时,现有控制算法难以对路面附着系数进行及时修正,并调整预测控制模型内的约束,进而导致车辆横向失稳.针对此种情况,提出一种考虑实时路面附着系数估计的横向跟踪控制策略,用于实现车辆横向轨迹跟踪.该算法针对路面附着系数未知的工况,利用车辆当前横纵向加速度、横摆角速度、前轮转角等状态量,通过扩展卡尔曼滤波预测路面附着系数后,再对控制模型中的侧偏角约束量进行实时调整,以保证车辆在未知路面工况下的行驶安全,使车辆跟随预期轨迹行驶.实验表明,将扩展卡尔曼滤波法与模型预测控制结合的控制算法具有可行性,且有效提高了车辆在不同附着系数路面行驶时横向轨迹跟踪的稳定性及鲁棒性.
Abstract
The traditional model predictive control method is used to track the vehicle trajectory,the road adhesion coefficient in the model is often fixed as the empirical value of the current road.When the vehicle is driving on unknown road,it is difficult to adjust the inboard constraint of predictive control model in real time,and it is difficult to correct the road adhesion coefficient in time,which leads to vehicle lateral instability.In this paper,a model predictive control algorithmconsidering real-time road adhesion coefficient is proposed to realize trajectory tracking.Under the condition of unknown road friction coefficient,the algorithm predicts the road adhesion coefficient through the vehicle's current transverse and longitudinal acceleration,yaw rate,front wheel angle and other state variables,and adjusts the yaw angle constraint in the control model in real time.Even under the unknown road surface,it can provide redundant protection for vehicle control and ensure the vehicle to follow the expected trajectory.
关键词
模型预测控制/扩展卡尔曼滤波/路面附着系数预测/轨迹跟踪控制Key words
model predictive control/extended Kalman filter/prediction of pavement adhesion coefficient/trajectory tracking control引用本文复制引用
基金项目
北京信息科技大学"勤信人才"培育计划项目(QXTCPA201903)
北京信息科技大学"勤信人才"培育计划项目(QXTCPB201901)
促进内涵发展科研水平提高项目(2020KYNH112)
出版年
2024
国家科技期刊平台
NSTL
万方数据