ES-MRAC: A New Paradigm For Adaptive Control

We develop a method for the model reference adaptive control (MRAC) of LTI systems via Extremum Seeking (ES). Our proof of global asymptotic tracking enables design of the adaptive controller to satisfy averaging requirements, and convergence of tracking error. Our method is novel, with the additional advantage that no perturbations need be added to the reference trajectory and no a priori knowledge of parameter signs is needed. We illustrate our results for a simulated second order system.