欢迎访问《汽车安全与节能学报》，
汽车安全与节能学报

汽车安全与节能学报 ›› 2024, Vol. 15 ›› Issue (1): 129-136.DOI: 10.3969/j.issn.1674-8484.2024.01.014

• 智能驾驶与智慧交通 • 上一篇    

考虑参数估计的MPC算法的商用车车道保持控制

赵崇钦(), 景晖(), 王刚, 冯焕秦, 刘夫云   

  1. 桂林电子科技大学 机电工程学院，桂林 541004，中国
  • 收稿日期:2023-08-01 修回日期:2023-09-26 出版日期:2024-02-29 发布日期:2024-02-29
  • 通讯作者: *景晖，研究员。E-mail：jinghui@guet.edu.cn。
  • 作者简介:赵崇钦（1995—）， 男（汉），广西，硕士研究生。E-mail：463810664@qq.com。
  • 基金资助:
    国家自然科学基金项目(52262052);广西创新驱动重大专项项目(桂科AA22372)

Lane-keeping control for commercial vehicles with an MPC algorithm considering parameter estimation

ZHAO Chongqin(), JING Hui(), WANG Gang, FENG Huanqin, LIU Fuyun   

  1. School of Mechanical and Electrical Engineering, Guilin University of Electronic Technology, Guilin 541004, China
  • Received:2023-08-01 Revised:2023-09-26 Online:2024-02-29 Published:2024-02-29

RichHTML

6

PDF

16

摘要：

设计了一种考虑参数估计的模型预测控制（MPC）算法的、智能辅助驾驶的商用车的车道保持算法，对于难以直接测量的质量和横向速度进行估计。建立车辆动力学模型和状态误差方程，通过扩展Kalman滤波（EKF）和递推最小二乘法（RLS）分别对车辆的横向速度、质量进行估计。基于估计得到的车辆参数，设计MPC车道保持控制器。构建硬件在环（HIL）仿真平台，设置不同的测试工况对车道保持算法进行了验证。结果表明：与普通MPC相比，在偏移回正工况中，车辆纠偏消耗的时间减少28.6%，并且超调量更小；高速路工况的横向位置偏差的均方根误差减小了4.2 cm。该方法提升了纠偏能力和跟踪精度，降低了传感器成本。

关键词: 智能辅助驾驶, 车道保持, 参数估计, 模型预测控制（MPC）, 递推最小二乘法（RLS）, 扩展Kalman滤波（EKF）

Abstract:

A lane-keeping algorithm was designed with the model predictive control (MPC) algorithm for commercial vehicles equipped with intelligent assisted driving. This algorithm took account parameter estimation and was capable of estimating mass and lateral velocity, which were difficult to directly measure. The extended Kalman filter (EKF) and recursive least squares (RLS) were used to estimate the lateral velocity and the mass of the vehicle, respectively. An MPC lane-keeping controller based on the estimated parameters was designed. A hardware-in-the-loop (HIL) was constructed. Different test conditions were established to verify the lane keeping algorithm. The results show that compared with the ordinary MPC, the time for vehicle correction is reduced by 28.6 % and the overshoot is smaller in the offset return condition. In the highway condition, the root mean square of the lateral error is reduced by 4.2 cm. At low sensor costs, the correction ability and tracking accuracy are improved.

Key words: lane-keeping, parameter estimation, model predictive control (MPC), recursive least squares (RLS), extended Kalman filter (EKF)

中图分类号: