Risk-sensitive filtering and parameter estimation

This paper investigates the use of risk-senstive filtering for state and parameter estimation in systems with model uncertainties. Modelling uncertainties arise from imperfectly known input process and noise characteristics as well as system model errors such as uncertain or time varying parameters of the system description. No new convergence results are given in this paper but simulation examples demonstrate that, in some situations, risk-sensitive filtering and estimation techniques allow for system uncertainties better than optimal techniques such as Kalman filterin