About a Possible High Poles Observer Design Instead of High Gain for Triangular Nonlinear Systems

This paper is about observer design for a class of nonlinear systems admitting a possible solution via the well-known high-gain technique. Here instead, a design based on full pole placement for the linear part of the observer is discussed. In particular it is shown that exponential convergence of the observation error is guaranteed in spite of nonlinearities, for poles chosen large enough, with observer gains which can be lower than in a standard high gain approach. A focus is given on systems up to order three, and a discussion on measurement noise effect is provided. Simulation examples are included to illustrate the conclusions.