Jochen Trumpf's talk on A Non-Linear Internal Model Principle for Observers
Title: A Non-Linear Internal Model Principle for Observers
Time, Date, Location: 11:00, Friday, 16 December, 2022, Brian Anderson Building Seminar Room
Abstract: We show that any asymptotic observer for a non-linear kinematic system on a differentiable manifold contains a full internal model of the plant. The result follows from an abstract, set-theoretic approximate internal model principle that generalises the known results for classical LTI systems, behavioural systems, and systems described by injective co-generator signal modules (Oberst’s algebraic analysis approach to system theory). At the core of the theory is the use of partially ordered set (poset) theory to generalise the classical LTI system notion of the anti-stabilizable part of a system. Most of the results in this talk were presented at the MTNS 2022 earlier this year.
The work is done in collaboration with Johannes Nuessle