汽车工程 ›› 2024, Vol. 46 ›› Issue (6): 985-994.doi: 10.19562/j.chinasae.qcgc.2024.06.005

• • 上一篇    下一篇

自动驾驶汽车避撞极限研究

王国栋1,2,刘立1,孟宇1(),杜海平2,白国星1,顾青1   

  1. 1.北京科技大学机械工程学院,北京 100083
    2.伍伦贡大学电子计算机与通信工程学院,伍伦贡 2522
  • 收稿日期:2023-11-23 修回日期:2024-01-10 出版日期:2024-06-25 发布日期:2024-06-19
  • 通讯作者: 孟宇 E-mail:myu@ustb.edu.cn
  • 基金资助:
    国家自然科学基金(52202505);中国国家留学基金(202206460040);中央高校基本科研业务费专项资金(FRF-IC-20-02);中国博士后科学基金(2022M710354)

Research on Collision Avoidance Limit of Autonomous Vehicles

Guodong Wang1,2,Li Liu1,Yu Meng1(),Haiping Du2,Guoxing Bai1,Qing Gu1   

  1. 1.School of Mechanical Engineering,University of Science and Technology Beijing,Beijing  100083
    2.School of Electrical,Computer and Telecommunications Engineering,University of Wollongong,Wollongong  2522,Australia
  • Received:2023-11-23 Revised:2024-01-10 Online:2024-06-25 Published:2024-06-19
  • Contact: Yu Meng E-mail:myu@ustb.edu.cn

摘要:

精确计算不同避撞控制策略的极限避撞距离是自动驾驶汽车避撞决策与控制的基础。为厘清差动制动控制对避撞距离的影响,探究转向和差动制动集成控制的极限避撞能力,提高极限避撞距离的计算精度,本研究基于非线性车辆集成动力学和最优控制理论提出一种自动驾驶汽车极限避撞距离计算方法。首先,建立了非线性 7自由度车辆动力学模型和复合滑移工况的Pacejka轮胎模型。进一步地,基于上述模型构建了极限避撞距离求解问题,并将其转化为最优控制问题。然后,设计了高斯伪谱法将最优控制问题转化为非线性规划问题并求解。最后,分析了转向控制、制动控制、转向和制动集成控制、转向和差动制动集成控制的极限避撞距离,并与基于质点模型计算和CarSim测试的结果进行了对比。结果表明:转向和差动制动集成控制能够进一步减少自动驾驶汽车的避撞距离,显著提高其避撞能力;本研究所提方法能够显著提高极限避撞距离的计算精度和避撞决策结果的可靠性。

关键词: 自动驾驶汽车, 车辆集成动力学, 避撞控制, 避撞极限, 高斯伪谱法

Abstract:

The precise computation of the limit collision avoidance (CA) distance for various collision avoidance control strategies is the basis for collision avoidance decision-making and control of autonomous vehicles. To clarify the impact of differential braking control on CA distance, explore the limit collision avoidance capability of steering and differential braking integrated control, and improve the calculation accuracy of the limit CA distance, in this study, a method for calculating the limit CA distance of autonomous vehicles based on nonlinear integrated vehicle dynamics and optimal control theory is proposed. Firstly, a nonlinear 7-degree-of-freedom integrated vehicle dynamics model and a Pacejka tire model with combined slip conditions are established. Subsequently, based on the established models, a limit CA distance solution problem is constructed, which is then transformed into an optimal control problem. Next, the Gauss Pseudospectral Method (GPM) is designed to convert the optimal control problem into a nonlinear programming problem and solve it. Finally, the limit CA distances for steering control, braking control, steering and braking integrated control, and steering and differential braking integrated control are analyzed, and compared with the limit CA distances calculated based on the particle model and that tested by CarSim. The results show that the integrated control of steering and differential braking can further reduce the CA distance and significantly improve the collision avoidance capability of autonomous vehicles. The method proposed in this study can significantly improve the calculation accuracy of limit CA distance and the reliability of CA decision results.

Key words: autonomous vehicles, integrated vehicle dynamics, collision avoidance control, collision avoidance limit, Gauss pseudospectral method