驾驶人行为不确定性对车辆轨迹预测的影响分析

Analysis of the influence of driver behavior uncertainty on vehicle trajectory prediction

针对驾驶行为不确定性影响车辆轨迹预测精度的问题,采用预瞄控制方法和滑模控制方法,分析了不同行驶车速与工况下车辆轨迹预测误差。结果表明:预瞄控制方法和滑模控制方法能够有效处理驾驶人超车、避障和转向行为的外部干扰,将轨迹预测误差控制在0.5 m以内,但高速工况下的轨迹误差明显大于低速工况下的轨迹误差;在小曲率转向工况下,基于单点预瞄驾驶人视觉轨迹预测误差小于基于双点预瞄驾驶人视觉轨迹预测误;在大曲率转向工况下,基于单点预瞄驾驶人视觉轨迹预测误差大于基于双点预瞄驾驶人视觉轨迹预测误。研究结果为驾驶人视觉轨迹预测策略的制定及车辆不同行驶工况下驾驶人行为不确定性的处理提供了理论依据。

Aiming at the problem that the uncertainty of driving behavior affects the accuracy of vehicle trajectory prediction,a predictive control method and a sliding mode control method are used to analyze the vehicle trajectory prediction error under different driving speeds and working conditions.The trajectory prediction accuracy can be improved by eliminating the deviation.On one hand,the preview control method is used to deviation between the current moment and the next time is estimated by predicting trajectory value,vehicle velocity,and yaw angle.Then it is used to correct driving steering.On the other hand,the sliding mode controller is adopted to amend the steering wheel angle,which can overcome the uncertainty of the driver's behavior.So vehicle safety is enhanced furtherly.The research results show that the preview control method and the sliding mode control method can effectively deal with the external disturbance of the driver's overtaking,obstacle avoidance,and steering behavior,control the trajectory prediction error within 0.5 m,achieve the purpose of improving the vehicle trajectory prediction accuracy,and enhance the vehicle trajectory prediction accuracy.It improves the driving stability of the driver and the safety of the vehicle.However,the trajectory error under high-speed conditions(90 km/h)is significantly larger than that under low-speed conditions(25 km/h);under small curvature steering conditions,the prediction error of the driver's visual trajectory based on the single-point preview is smaller than that based on double-point preview.Under the large curvature steering condition,the prediction error of the driver's visual trajectory based on the single-point preview is larger than that based on the double-point preview.The prediction model is stronger than the two-point preview driver model in terms of timeliness,handling stability,and anti-interference ability.The research results provide a theoretical basis for the formulation of the driver's visual trajectory prediction strategy and the handling of the driver's behavior uncertainty under different vehicle driving conditions.