Research Overview
The research in the lab focuses on development of control theoretic methods in the following areas:
- optimal and learning-based control for nonlinear systems,
- analysis of uncertain, dynamical systems using set-based methods,
- hierarchical, optimal and fault-tolerant control,
- formal verification of control systems.
The research in the lab has a strong focus on controller development for the full cycle starting from system modelling to the end application. The particular topics also include advanced techniques of data analysis and classification along with model uncertainty assessment.
Special attention is paid to complex and hierarchical systems, where consistency between control and subroutines for different system layers is of particular interest. Special attention is placed on
controlled environment agriculture,
fuel cell systems, and
trains as application areas.
Concrete control methods in turn include model predictive control, reinforcement learning, adaptive dynamic programming and other optimal control methods, with a special focus on learning aspects.
Mathematical proofs and analysis of e.g. robustness properties of control methods in the presence of uncertainties is a part of virtually every new control method developed in the lab.
We also try to bridge the gap between theory and computer implementation by explicitly dealing with computational uncertainty related to the imperfections of controller implementation in digital and analog devices. For this sake, the lab works on a variety of mathematical methods which explicitly incorporate the said uncertainty and subsequently aim at developing a basis of formally correct and automated controller extraction.
