Comments on ``Gain scheduling dynamic linear controllers for a nonlinear plant'' (Q1294983)

These comments refer to the procedure used for choosing the realization of a gain-scheduled controller in two papers, authored by (i) \textit{D. A. Lawrence} and \textit{W. J. Rugh} [ibid. 31, No. 3, 381-390 (1995; Zbl 0825.93192)] and (ii) \textit{I. Kaminer}, \textit{A. M. Pascoal}, \textit{P. P. Khargonekar} and \textit{E. E. Coleman} [ibid. 31, No. 8, 1185-1191 (1995; Zbl 0839.93037)], respectively. The discussed procedure requires that the linearization of the gain-scheduled controller, at an equilibrium-operating point, should correspond to the associated member of the family of linear controllers. The authors of the comments claim that many different controller realizations satisfy the local linear equivalence condition (even if the realizations are not equivalent) and can exhibit quite different dynamics in the vicinity of an equilibrium operating point. An example is given to support these statements and to illustrate the fact that the local linear equivalence condition can also lead to the imposition of restrictive constraints on the controller states and inputs which are not a priori necessary. On the other hand, in their reply, the authors of paper (i) bring further arguments to show that the criticized nonuniqueness aspects of the gain scheduling conditions do not represent a disadvantage of the proposed procedure. Moreover, they consider that the example provided by these comments (where the restrictive constraints on the controller states and inputs can be avoided) is not relevant, since no rigorous proof is presented to cover the general case.