Action Group on Probabilistic and Randomized Methods

Abstract

This site is maintained by the Virtual Action Group on Probabilistic and Randomized Methods in Control, which is part of the IEEE Control Systems Society technical committee on Computational Aspects of Control System Design (CACSD).

It is intended to collect, sort and present available and relevant information on probabilistic methods for the analysis and control of dynamic systems, with special focus on the computational aspects, such as randomized algorithms and other optimization techniques.

Introduction

A well-established approach to robust performance analysis and design of control systems is based on a purely deterministic unknown-but-bounded description of the uncertainty. Fundamental results in this direction have been obtained in the last thirty years. Recently, however, this deterministic - or worst-case - approach also showed some inherent limitations, that can be resumed in the fundamental tradeoff between computational complexity and "conservatism". In fact, exhaustive search through solution or parameters spaces get rapidly out of hand and can be unworkable for real world problems. Output feedback stabilization of a plant and closed-loop stability of plant-controller pairs when the plant is subjected to structured perturbations exemplify this kind of issues. Also, from a more philosophical perspective, a worst-case design, even if it could come at a cheap computational cost, may not be a desirable design in practice, since it contemplates all possible uncertainty scenarios, including those that are extremely unlikely to happen. This motivated the introduction of a probabilistic approach to robustness analysis and design, and the related computational tools, such as randomized algorithms. The starting point of this approach is to introduce a probabilistic description of the uncertainty, and to provide probabilistic assessments on the satisfaction of design specifications. In other words, the performance levels are not guaranteed against all possible uncertainty instances, but only for most (the most probable) of them.

 

Goals

The site compiles researchers who are active in the field, research projects that are underway, links to a number of selected papers and software, and it provides news about conferences, workshops, special issues, and the like. Any suggestions or comments to improve this site are more than welcome. For contributions or comments please contact the AG Chair.