Repetitive Learning Control For Global Trajectory Tracking Of Robot Manipulators

Robot manipulator is a main class of highly complex and strong coupling nonlinearsystem. With the development of industrial automation, high precision control problemof robot manipulator has become a hot research field. Based on the observation that therobot manipulator often performs repeated movement, repetitive learning controlstrategy is one of the methods to address the high precision tracking problem of robotmanipulator.In this paper, we provide three kinds of repetitive learning control methods. First, asimple linear repetitive learning controller is proposed, which consists of a PD actionand a learning-based feedforward compensation term. The faster response and highertracking precision is obtained over PD control without increased torque. Second, a newnonlinear sliding mode repetitive learning control is constructed which is formulatedwith a class of function with the characteristics of "enlargement of small error". Third, anew adaptive repetitive learning control is proposed by using the adaptive controltechnology to overcome the uncertainty of the system. The higher tracking performanceis also obtained. By the Lyapunov’s direct method and Barbalat’s lemma, the proposedcontrols are proven to ensure global asymptotic tracking. The simulation results on atwo degree-of-freedom robot illustrate the effectiveness and improved performance ofthe proposed controllers.