“…In this context, an alternative consists in aiming at improving the estimation performance of a given observer. To the best of the authors' knowledge, existing works in this direction either concentrate on specific classes of systems (see e.g., [4]- [6] for linear systems or e.g., [7]- [12] in the context of high-gain observers) or on specific properties like robustness to measurement noise (see, e.g., [13], [14]) or the reduction of the undesired effect of the peaking phenomenon (see e.g., [15] for a general approach and [8], [10], [12] for specific solutions in the context of high-gain observer). An exception is [16], where two observers designed for a general nonlinear system are "united" to exploit the good properties of each.…”