七、学习控制在机器人控制中,操作手跟踪预定的轨迹常是控制的主要任务。这时,既使控制系统的跟踪性能很好也会出现轨迹偏差。为了清除这种偏差,日本学者Arimoto等人提出采用学习控制方法。其基本思想:在有限的时间区域[O,T]内,预定的轨迹y_d(t)表示控制系统的希望输出。通过机器人的反复操作和训练,可以找出改善控制输入v(t)的一种学习方式,即利用前一次操作数据和轨迹偏差,改善下一次控制输入,最后求得 Seven, learning control In the robot control, the operator tracks the scheduled trajectory is often the main task of control. At this time, trajectory deviation occurs even if the tracking performance of the control system is good. In order to eliminate this deviation, Japanese scholar Arimoto et al proposed the use of learning control method. The basic idea is that within a limited time zone [O, T], the predetermined trajectory y_d (t) represents the desired output of the control system. Through repetitive operation and training of the robot, a learning method of improving the control input v (t) can be found, which is to improve the next control input by using the previous operation data and track deviation, and finally obtain