Organized iterative learning control for trajectory tracking

In trajectory tracking applications where a single trajectory is followed many times, Iterative Learning Control (ILC) is used to improve tracking performance by compensating for controller lag and disturbances that are repetitive across iterations. These tracking errors are not necessarily found throughout the iteration and may even be sufficiently learned after only a few iterations. Influenced by segmented and multirate control, this paper presents a new ILC algorithm which reduces how often the ILC input signal is updated as learning progresses. Portions of the signal where sufficient learning has occurred are divided and approximated as constant based on where the magnitude of the input is small and is slowly changing. Organized ILC is compared to the p-type ILC formulation and is shown to perform just as well as the All cycle learning. During sections of constant velocity, the organized ILC quickly compensates for the error as does the p-type ILC. In portions where tolerances are satisfied, the organized ILC begins partitioning and approximating the input signal and is shown to significantly reduce the number of times the input signal is updated Measurement noise is also introduced and the RMS of the error signal for each iteration is compared. The organized ILC is shown to handle measurement noise significantly better than p-type ILC.