Asymptotic tracking in a class of uncertain nonlinear systems via learning-based inversion

A novel learning approach is presented to solve asymptotic state-tracking problem for a class of uncertain nonlinear systems, where the tracking-error dynamics are described by a set of time-periodic nonlinear differential equations. The proposed approach utilizes the specific property that the closed-loop system tends to oscillate in steady state and, thereby, extends in a natural manner the idea of the well-known iterative learning control to infinite-time asymptotic tracking problems. Existence of the steady-state oscillation and convergence of the proposed iterative update scheme is developed through rigorous theoretical analysis. Simulation results are also presented.