ILC is suitable in situations where a *repetitive task* is to be performed multiple times, and disturbances acting on the system are also repetitive and predictable but may be unknown. Multiple versions of ILC exists, of which we support a few that are listed below. When ILC iterations are performed by running experiments on a physical system, ILC resembles episode-based reinforcement learning (or adaptive control), while if a model is used to simulate the experiments, we can instead think of ILC as a way to solve optimal control problems (trajectory optimization).
0 commit comments