汽车工程 ›› 2024, Vol. 46 ›› Issue (7): 1197-1207.doi: 10.19562/j.chinasae.qcgc.2024.07.007

• • 上一篇    

基于学习的无人驾驶车辆模型预测路径跟踪控制研究

韩陌1,何洪文1(),石曼1,刘伟2,曹剑飞3,吴京达4   

  1. 1.北京理工大学,高端汽车集成与控制全国重点实验室,北京 100081
    2.上海友道智途科技有限公司,上海 200438
    3.北京空间飞行器总体设计部,北京 100094
    4.香港理工大学,香港 999077
  • 收稿日期:2024-02-07 修回日期:2024-03-22 出版日期:2024-07-25 发布日期:2024-07-22
  • 通讯作者: 何洪文 E-mail:hwhebit@bit.edu.cn
  • 基金资助:
    国家自然科学基金(52172377)

Research on Learning-Based Model Predictive Path Tracking Control for Autonomous Vehicles

Mo Han1,Hongwen He1(),Man Shi1,Wei Liu2,Jianfei Cao3,Jingda Wu4   

  1. 1.Beijing Institute of Technology,National Key Laboratory of Advanced Vehicle Integration and Control,Beijing 100081
    2.UTOPILOT,Shanghai 200438
    3.Beijing Institute of Spacecraft System Engineering,Beijing 100094
    4.The Hong Kong Polytechnic University,Hong Kong 999077
  • Received:2024-02-07 Revised:2024-03-22 Online:2024-07-25 Published:2024-07-22
  • Contact: Hongwen He E-mail:hwhebit@bit.edu.cn

摘要:

针对无人驾驶车辆路径跟踪控制问题中预测模型准确性与计算成本平衡制约问题,本文提出了一种基于学习的模型预测(learning-based model predictive control, LB-MPC)路径跟踪控制策略。建立了车辆2自由度单轨动力学模型,深入分析了其与IPG TruckMaker模型单步响应误差随车速、踏板开度及前轮转向角的变化规律,设计了误差数据集构建和滚动更新方法,基于高斯过程回归(Gaussian process regression, GPR)建立了误差拟合模型,对标称单轨模型进行实时误差补偿修正。将误差修正模型作为预测模型,设计了路径跟踪代价函数,构建了二次规划优化问题,提出了一种基于学习的模型预测路径跟踪控制架构。基于IPG TruckMaker & Simulink联合仿真平台与实车平台,验证了所提GPR模型误差修正与LB-MPC路径跟踪控制策略的实时性与有效性。结果表明,相较于传统模型预测(model predictive control, MPC)路径跟踪控制策略,所提LB-MPC策略路径跟踪平均误差降低了23.64%。

关键词: 路径跟踪, 车辆模型误差分析, 高斯过程回归, 模型预测控制

Abstract:

For the trade-off between prediction model accuracy and computational cost for path tracking control of autonomous vehicles, a learning-based model predictive control (LB-MPC) path tracking control strategy is proposed in this paper. A two-degree-of-freedom single-track vehicle dynamic model is established, and an in-depth analysis is conducted on its step response error with respect to variation in vehicle speed, pedal position, and front wheel steering angle compared to the IPG TruckMaker model. Methods for constructing error datasets and receding horizon updates are designed, and the Gaussian process regression (GPR) is employed to establish an error-fitting model for real-time error compensation and correction of the nominal single-track model. The error correction model is utilized as the prediction model, and a path tracking cost function is designed to formulate a quadratic programming optimization problem, proposing a learning-based model predictive path tracking control architecture. Through joint simulation using the IPG TruckMaker & Simulink platform and real vehicle experiments, the real-time performance and effectiveness of the proposed GPR error correction model and LB-MPC path tracking control strategy are verified. The results show that compared to the traditional model predictive control (MPC) path tracking control strategy, the proposed LB-MPC strategy reduces the average path tracking error by 23.64%.

Key words: path tracking, vehicle model error analysis, Gaussian process regression, model predictive control