Iterative Learning Control

摘要

Abstract A theory of iterative learning control for refinement of motions of robotic systems is presented, together with simulation results. It is shown that the passivity of such non-linear mechanical systems plays a key role in the ability to acquire a desired and skilled movement through repeated practice. This iterative learning scheme can be also applied to robot dynamics under holonomic tool tip constraint. When fluctuations and measurement noise exist, the introduction of a forgetting factor in the learning update law becomes crucial in the convergence of trajectory tracking errors.