Synthesis and Implementation of a Direct Learning Control

This paper considers the problem of synthesizing a direct learning control (DLC) law for a class of nonlinear systems which perform given tasks repetitively. First, it is shown that DLC can generate the desired control input directly from prestored control input profiles which were learned for other output trajectories. It is also illustrated by simulations and experiments that the DLC method can be effectively applied to the tracking control of a SCARA robot manipulator using the information on the relative degree of the robot system.