Design of observer-based robust stabilizing controllers

Abstract This paper is concerned with the problem of designing stabilizing control laws for linear systems with time-varying uncertainty. Since the entire state cannot be measured, the controller must operate on some estimate of the state. The approach used is to first obtain a linear robust controller based on the assumption that the complete state is available. Then an estimated state is generated via an observer and this estimate used in the implementation of the controller. Examples indicate that the use of an observer usually reduces the magnitude of the tolerable uncertainty. The amount of this reduction depends on the choice made for the measured state variables.