基于人机共享和分层控制的车道偏离辅助系统

Human-machine Sharing and Hierarchical Control Based Lane Departure Assistance System

针对车道偏离辅助系统转向控制中的人机协同问题,提出一种人机共享的决策与控制方法。考虑驾驶员转矩、道路曲率及纵向车速,设计模糊控制器确定虚拟车道边界宽度,再根据车轮是否超出虚拟边界进行辅助控制决策。在此基础上根据车-路位置关系建立车辆预瞄处偏差的动态模型,设计LPV/H∞控制器跟踪车道中心线决定期望前轮转角。应用滑模观测器估计转向阻力矩,设计考虑驾驶员转矩和转向阻力矩的二阶滑模控制器决定辅助转矩,再由主权分配模块进一步对人机控制主权进行分配,实现辅助系统与驾驶员之间的共享控制。在veDYNA/Simulink联合仿真平台上对提出的控制方法进行仿真试验,仿真结果表明所提出方法在各种车速下均能有效避免车辆偏离车道,减少控制器对驾驶员的干预,人机冲突较小。建立基于veDYNA/LabVIEW的车道偏离辅助控制系统试验台架,对该方法进行硬件在环仿真试验,其结果表明该方法能有效辅助驾驶员通过复杂道路,获得较好地人机协调性能。

According to the human-machine coordination of steering control for lane departure assistance system（LDAS）, a human-machine shared decision-making and control method is proposed. Accounting for the driver torque, road curvature and longitudinal speed, the virtual strip width is determined using a fuzzy controller and assistance control is based on whether the wheels are beyond the virtual strip. Then, according to the position relation between vehicle and lane, the dynamic model of vehicle lateral deviation at look-ahead point is established. The LPV/H∞ controller is designed to tracking center-line of the lane and determining the desired steering angle. A sliding mode observer with unknown inputs is designed to estimating the road reaction torque. A second-order sliding mode controller which take into account driver torque and road reaction torque is used to generate assist torque. The human-machine control authority is shifted by authority distribution module to achieve shared steering control between assistance system and driver. The proposed assistance system is evaluated in veDYNA/Simulink, simulation results showed that the proposed method can effectively avoid lane departure at different speeds, minimize controller intervention and human-machine conflict. A test platform for LDAS is established, which is based on veDYNA/LabVIEW. The proposed method is experimentally validated with driver and hardware in the loop, the result showed that this method can effectively assist the driver through the complicated lane and good results were obtained using a criterion for human-machine cooperation.