25-29 Nov 2018, IIT Bombay

Overview

Control techniques based on algorithmic selection of actions derived from optimization of suitable performance indices are increasingly gaining prominence in today’s world. Predictive control techniques are at the forefront of these developments. While the deterministic and robust versions of predictive control techniques are largely standard today, modern applications have spawned newer areas such as economic and financial predictive control, that are rapidly gaining prominence. Concurrently, driven by a world-wide thrust towards data-driven paradigms, stochastic predictive control techniques are also becoming increasingly popular.

However, the technical tools in these two domains appear to be significantly disjoint from each other. Indeed, on the one hand, even though several computationally viable tools are available in the deterministic and robust predictive control community, they tend to under-perform in the stochastic setting. On the other hand, while there exist several strong existential results in the stochastic community, the set of computational tools available today is relatively sparse. Moreover, it appears that most researchers working in one of these two areas tend to work in isolation and/or ignorance of the technical tools available with the other area.

This workshop seeks to bring together some of the key architects behind the latest innovations in both deterministic and stochastic predictive control and engage in technical discussions to cross-inform and cross-fertilize their respective areas while staying focussed on the crucial strength of predictive control: algorithmic synthesis techniques for constrained control.

Organizers

• Daniel Quevedo, Paderborn University, Germany
• Debasish Chatterjee, IIT Bombay, India
• Vivek Borkar, IIT Bombay, India