Learning Control in Practice: Novel Paradigms for Industrial Applications

摘要

This paper presents a number of solutions to issues with learning control in practice. We use RoFaLT, a freely-available, model-based iterative learning control (ILC) tool for nonlinear systems, which implements an optimization-based two-step approach. We augment it with concepts to improve robustness, convergence speed, and avoid high computational loads online. These concepts are illustrated on an electromechanical set-up with slider-crank mechanism.