Nonlinear iterative learning by an adaptive Lyapunov technique

Summary form only given. The iterative learning problem is considered from an adaptive control viewpoint. We consider a wide variety of systems (e.g. linear, matched, strict feedback, output feedback) and show that by simple modifications to wide classes of adaptive designs, we can utilize the adaptive designs to solve the ILC problems. The main difference between these adaptive ILC designs and their conventional ILC counterparts is that the ´learning´ is achieved during the passes, as opposed to the standard designs which only allow learning computations to occur at the end of each pass. The most obvious advantage is that the designs are robust to global parameter variations. In contrast to many other nonlinear ILC designs, we have no need to make restrictive global Lipschitz assumptions, nor do we restrict attention to systems with matched uncertainty