Dynamic compensation in robust stabilization using observers

This paper is concerned with the stabilization of uncertain linear systems via observer-based controllers. The motivation comes from earlier work (see [6] and [7]) where sufficient conditions are given for uniform asymptotic stability of the combined plant-observer-controller. In this paper, we describe an augmented dynamical system whose stability via an observer-based controller implies likewise for the original plant. While the given system may have uncertainties in the matrices describing the model, input connection, output distribution and direct transmission, the augmented system only contains uncertainties in its model matrix. The advantages of "lumping" the uncertainty into the model matrix are also discussed.