A robust controller interpolation design technique

In this paper, a robust controller interpolation technique is investigated and applied to an experimental test bed. Controller interpolation describes the process of switching or blending among a family of controllers. The proposed approach allows the individual controller designs to be decoupled by designing the controller interpolation after the family of controllers is designed. In contrast, the 'self-scheduled' linear parameter varying techniques simultaneously design the family of controllers and the controller interpolation, thereby coupling the individual controller designs to one another. The presented controller interpolation framework uses the classic controller blending framework augmented by stabilizing compensators. The synthesis conditions for the stabilizing compensators maximizing robust stability is stated in terms of a set of bilinear matrix inequalities (BMI). In the motivating example, the stabilizing compensators are designed by using a search that is locally minimizing over the robustness metric.