驾驶机器人车辆多新息动态转向力矩补偿

Multi-Innovation Based Dynamic Steering Torque Compensation for Driving Robot Vehicle

为了减小驾驶机器人车辆长期自动驾驶过程中转向性能下降带来的影响,提出了一种基于多新息的驾驶机器人车辆动态转向力矩补偿方法。构建了车辆动力学模型和驾驶机器人车辆动力学模型;建立了以路径曲率及车速为输入、方向盘转向角为输出的驾驶机器人车辆转向性能离线自学习模型;建立了以方向盘角速度、角加速度及车轮转角为输入,转向机械手驱动力矩为输出的受控自回归在线辨识模型,并运用遗忘因子多新息最小二乘方法进行参数辨识,将迭代计算过程中的标量新息扩展为向量新息,提高了驾驶机器人车辆转向性能参数的辨识精度;驾驶机器人车辆自动驾驶过程中,利用离线自学习模型和转向机械手动力学方程计算出转向电机输出力矩,加上反馈回来的驱动力矩误差,实现对驾驶机器人车辆转向力矩的在线动态补偿。仿真与试验结果对比表明:所提方法辨识的转向力矩误差在0.1N·m以内,跟踪目标路径的横向位移偏差小于0.2m;所提方法有效减小了驾驶机器人车辆转向性能下降造成的影响。

To weaken the impact of steering performance degradation during long-term automatic driving,a dynamic steering torque compensation method for driving robot vehicle based on multi-innovation is proposed.The vehicle dynamic model and driving robot vehicle dynamic model are constructed,then an off-line self-learning model for the steering performance of driving robot vehicle is established,which takes the path curvature and the vehicle speed as input and the steering wheel angle as output.And a controlled autoregressive on-line identification model is established,which takes the angular velocity of steering wheel,angular acceleration and wheel angle as input and the driving torque of steering manipulator as output,and the parameters are identified with the forgetting factor multi-innovation least square method,the scalar innovation is extended to vector innovation in the iterative calculation to improve the identification accuracy of steering performance parameters of driving robot vehicle.In the automatic driving process of driving robot vehicle,the off-line self-learning model and steering manipulator dynamic equation are used to calculate the output torque of steering motor,and the fed-back driving torque error is added to realize the online dynamic compensation of steering torque of driving robot vehicle.A comparison between simulation and test results shows that the steering torque error identified by the proposed method is within 0.1 N·m,and the lateral displacement error of tracking the target path is less than 0.2 m.