In this thesis, a framework for the real-time control and performance recovery of highly dynamic systems is proposed and implemented. The most distinctive characteristic of the framework is that the computationally expensive tasks are realized in a distributed manner, hierarchically and functionally, thanks to the observer based realization of the Youla parameterization of all stabilizing controllers. Under this framework, algorithms referring to image processing and real-time performance recovery are presented.