Norm optimal iterative learning control based on a multiple model switched adaptive framework

Abstract

In this paper a prominent class of iterative learning control (ILC) algorithm is reformulated in the framework of estimation-based multiple model switched adaptive control (EMMSAC). The resulting control scheme uses a bank of Kalman filters to assess the performance of a set of candidate plant models, and the ILC update at the end of each trial is constructed using the plant model with smallest residual. Through exploitation of the powerful underlying EMMSAC framework, rigorous bounds are available to guarantee robust ILC performance without placing constraints on the form of uncertainty or control action.

This paper hence addresses current limitations in ILC approaches for uncertain systems which are typically highly restrictive in the form or magnitude of the uncertainty, employ prescribed controller forms, or, alternatively are heuristically motivated with no theoretical stability/performance guarantees. Experimental results from a highly relevant application of ILC in stroke rehabilitation are given to confirm the efficacy and scope of the framework.

More information

Identifiers

Catalogue record

Export record