In many real life situations, we must design a controller for a system whose parameters may vary, and vary widely. E.g., for an auto-piloted regime of the supersonic Concorde plane, one of the parameters varies from 0.02 to 0.2. Due to wide variations, we cannot simply neglect the terms quadratic in these variations and use linearized methods.
In many such applications (Concorde one of them), errors will be catastrophic, so, we need to guarantee stability.
In the paper under review, a new method of controller design is developed for such situations; this method is illustrated on the example of the Concorde.