An improved high gain observer for single-output uniformly observable systems

In this paper, we present an improvement of the classical high gain observer given in Gauthier et al (1992) for single-output uniformly observable systems. The improvement lies basically in the fact that the proposed observer incorporates the information of the nonlinearity of the system in the design strategy. Consequently, unlike the classical observer, the gain of the improved observer is function of the state estimate. A comparison between the proposed observer and the classical high gain one is made via a simulation example dealing with a flexible joint mechanism. It is shown that the improved observer is more robust with regards to measurement noise and present less transients oscillations. An extension of the design is also presented for a more general class of nonlinear systems.