Ivan Hasenohr, MAP5
Computer-assisted methods for rigorous description of reachable sets in constrained linear control problems
We consider a finite-dimensional linear control system subject to (convex, compact) constraints. The goal is to describe the reachable set, which involves providing evidence of the inclusion (or non-inclusion) of a target within the reachable set from a given initial condition and within a given time T.
The developed methods rely on convex analysis tools, specifically the computation of supporting hyperplanes to the reachable set. Numerical implementation introduces discretization and rounding errors, which can be controlled to provide rigorous proofs.
The presented results are the outcome of ongoing work with Camille Pouchol, Yannick Privat, Christophe Zhang, and Maxime Breden.