Robust Control

Given an uncertain dynamical system, robust control aims to synthesize a controller that reaches a goal set while avoiding collisions with unsafe sets on the way. The presence of uncertainties in the initial state and the disturbances require the development of sophisticated controller synthesis approaches -- the resulting controller has to provably ensure that all initial states reach the goal set regardless of the disturbances without causing collisions. We address a wide range of topics, including motion-primitive-based control, model predictive control, and the computation of safe terminal and invariant sets.

Research questions include

• How can we improve the scalability of invariant set computation?
• How can we leverage reachability analysis to synthesize formally safe controllers?
• How can we construct motion primitives for fast online matching in motion planning?

The developed control algorithms are then applied in our applications. Our research in reachability analysis is implemented in our toolbox AROC.