Online State Estimation Via Deterministic Optimal Control Strategies

The presentation is concerned with the Mortensen observer, also called Minimum Energy Estimator, designed for the online state estimation of disturbed dynamical systems based on incomplete measurements. After briefly recalling its derivation based on principles from optimal control we address the analytical challenges stemming from its dependence on the value function. The focus lies on a numerical realization of the estimator based on a data-driven polynomial approximation of the value function mitigating the curse of dimensionality. The talk concludes with a design for linear models subject to parametric system uncertainties.