汽车工程 ›› 2022, Vol. 44 ›› Issue (3): 412-425.doi: 10.19562/j.chinasae.qcgc.2022.03.014

所属专题: 底盘&动力学&整车性能专题2022年

• • 上一篇    下一篇

紧急避撞工况下驾驶员异常行为检测

赵治国(),李园,胡昊锐,王鹏,陈晓蓉   

  1. 同济大学汽车学院,上海  201804
  • 收稿日期:2021-10-14 修回日期:2021-12-06 出版日期:2022-03-25 发布日期:2022-03-25
  • 通讯作者: 赵治国 E-mail:zhiguozhao@tongji.edu.cn
  • 基金资助:
    国家自然科学基金(52172390)

Detection of Driver’s Abnormal Behavior Under Emergent Collision Avoidance Conditions

Zhiguo Zhao(),Yuan Li,Haorui Hu,Peng Wang,Xiaorong Chen   

  1. School of Automotive Studies,Tongji University,Shanghai  201804
  • Received:2021-10-14 Revised:2021-12-06 Online:2022-03-25 Published:2022-03-25
  • Contact: Zhiguo Zhao E-mail:zhiguozhao@tongji.edu.cn

摘要:

在紧急避撞场景下,驾驶员极易因慌乱和误判,产生异常驾驶行为,故提前检测出驾驶员的异常行为对于确保驾驶员自身和周边交通的安全有重要意义。为此,本文中提出了综合转向盘转角残差序列和相平面图判断的异常驾驶行为检测方法。首先,基于相平面法确定车辆稳定性边界;接着,建立基于模型预测控制的驾驶员模型,构造实际驾驶操作与驾驶员模型参考输入间的转向盘转角残差序列;最后,计算驾驶员异常行为因子和评价函数,实现了对车辆偏离预期行驶轨迹的异常行为检测。仿真和硬件在环试验结果表明,所提出的异常驾驶行为检测方法能最晚于车辆偏离轨迹0.5 s前检测出驾驶员异常行为。

关键词: 紧急避撞, 驾驶员异常行为检测, 残差, 异常行为因子, 评价函数

Abstract:

In the scene of emergent collision avoidance, driver is prone to abnormal driving behavior due to panic and misjudgment. Therefore, detecting the abnormal behavior of driver in advance is of great significance to ensure the safety of the driver himself and the surrounding traffic. In view of this, a detection method of abnormal driving behavior that integrates the steering wheel angle residual sequence and the phase plane diagram judgment is proposed in this paper. Firstly, vehicle stability boundary is determined by using phase plane method. Then, the driver model is established based on model predictive control and the steering wheel angle residual sequence between the actual driving operation and the driver model reference input is constructed. Finally, the abnormal behavior factor and evaluation function of the driver are calculated to detect the abnormal behavior of the vehicle deviating from the expected driving trajectory. The results of simulation and hardware-in-the-loop test show that the abnormal driving behavior detection method proposed can detect the abnormal behavior of driver at least 0.5 s before the vehicle deviates from the desired trajectory.

Key words: emergent collision avoidance, driver’s abnormal behavior detection, residual error, abnormal behavior factor, evaluation function