Open-loop worst-case identification of nonSchur plants

This paper presents an LMI based algorithm for deterministic worst-case identification of nonSchur plants in an open-loop setting. Contrary to other approaches dealing with this problem, the proposed technique does not require prior knowledge of a stabilizing controller. The main result of the paper shows that, as the information is completed, the identified model converges, in the ?₂-induced topology, to the actual plant. Additional results include upper bounds on the worst-case identification error on the finite horizon. The usefulness of the proposed approach is illustrated with a practical example arising in the context of robust visual tracking.