Augmented system approach to measurement-based robust tracking (Q1368026)

A class of linear time-invariant systems is considered where parametric uncertainties satisfy a quadratic inequality. The reference signal is supposed to be a polynom of a fixed degree. In the first case, the outputs required to track the reference are assumed to be obtainable from observations. In the second case, the tracking outputs are independent of the observations. For both cases, design procedures based on solving a set of linear matrix inequalities are shown. However, in the first case, the procedure involves a stabilization problem only and ensures zero steady-state error but in the second one, the proposed approach can be applied only to step references and results in bounded errors.