Efficient Linear MIMO Adaptive Inverse Control

Abstract Adaptive inverse control is an automatic control-system design method which "learns" over time how to control a particular dynamic system. This simple controller design approach can achieve very precise control of minimum- and nonminimum-phase systems, even with poor a priori knowledge of the plant dynamics. A problem with adaptive inverse control in its current state is that the learning algorithms for MIMO and nonlinear systems are very slow. This paper addresses some results concerning a new way to perform adaptive inverse control of linear MIMO systems which learns very quickly. An example is presented to demonstrate the dramatic speedup.