Observability of systems when measurements are not time-stamped

Eric Busvelle (LIS, Université de Toulon)

Mercredi 14 décembre, 11h, salle Coriolis (Galois).

Abstract. Usually, observers’ theory always assumes that the outputs correspond to correctly timestamped states (although variable delays are sometimes taken into account). In practice, this is not always the case. We will give some examples to illustrate the problem, including a practical case from a real robotic application, still under development. The extreme situation is the complete absence of timestamps.
We will propose a definition of observability when the measurements are not timestamped. We will justify this non-trivial definition with a few examples. We will then try to estimate the rarity of the observable systems despite the absence of timestamps on the outputs.