Iterative learning control for robot manipulators using the finite dimensional input subspace

This paper proposes a new type of iterative learning control algorithm for manipulators, which seeks the desired input in an appropriate finite dimensional input subspace. The method achieves perfect tracking for uncertain manipulators without using time derivative of tracking error signals in the learning law. The convergence condition of the learning process is derived, and the effectiveness of the proposed method is evaluated by experiments using a two-link direct drive arm.