Iterative learning control for nonlinear systems with input constraints and discontinuously changing dynamics

This paper discusses the implementation and application of an iterative learning control (ILC) algorithm for nonlinear systems with input constraints and discontinuously changing dynamics. The ILC approach consists of two steps: in the first step the nominal model of the plant is corrected based on the previous iteration's output, and in the second step the corrected model is inverted to track the reference. Both steps are formulated as nonlinear least squares problems with a sparse, banded structure, and solved using an efficient implementation of an interior point method called IPOPT. Due to this implementation high order systems and long data sets can be efficiently processed. The considered application is a trajectory tracking problem of a one degree-of-freedom robot arm carrying an object that is suddenly released during motion. Both the case where the exact time instant of release is known and unknown are considered.