Unified Output Feedback Design and Loop Transfer Recovery

This paper presents an entirely new observer design algorithm for loop transfer recovery (LTR). This algorithm has three advantages over other LTR design algorithms when exact LTR cannot be achieved (true for most of cases). First, it results in finite gain, instead of asymptotically large gain of other algorithms. Second, it is much simpler, more explicit and systematic. Third, it can perfectly unify LTR and output feedback design within only a set of linear equations. The first two advantages are valid for all design cases, while the third advantage is valid as long as the open loop system has more outputs than inputs. This algorithm is based on a recent breakthrough on observer eigenstructure assignment, which forms the entire freedom of observer design (for LTR). Our solution to this problem is uniquely able to fully use this design freedom.