Performance and robustness issues in iterative learning control

Closed-loop system robustness and performance in the presence of uncertainty are important issues in the design of feedback systems. Much of the recent work in control theory has led to the development of methods that rigorously address these issues. The motivation for the so-called iterative learning control systems is similar: iterative learning control systems can improve performance and reduce the effects of uncertainty. However, these issues usually receive only heuristic treatment in the design of learning controllers. The purpose of this paper is to extend the rigorous techniques for robust stability and performance to an iterative learning control architecture. The paper discusses precise definitions of robust convergence and performance for this architecture, and demonstrates the authors' initial analysis and design results.<<ETX>>