Recursive estimator for linear and nonlinear systems with uncertain observations

The closed form solution for directly assigning steady state estimation error covariances and their assignability conditions are derived for the linear case and upper bounds on the estimation error covariance are assigned for the nonlinear case.

Abstract

The state estimation problem with observations which may or may not contain a signal at any sample time is considered from a covariance assignment viewpoint. The closed form solution for directly assigning steady state estimation error covariances and their assignability conditions are derived for the linear case. For the nonlinear case, upper bounds on the estimation error covariance are assigned. Examples are given for illustration in which the robustness of the proposed schemes are assessed.