A new self-tuning controller to solve the servomechanism problem

The design of "universal regulators" which can adaptively stabilize unknown linear time-invariant (LTI) systems has been a recent topic of interest. Martensson[1] has shown that the order of any LTI stabilizing regulator is sufficient a priori information in order to achieve adaptive stabilization of a minimal LTI plant. A more practical stabilizing regulator which requires the same a priori information about a stabilizable and detectable plant is given in [2]. In this paper we use the ideas of [2] to design a controller to solve the servomechanism problem for an unknown system under very mild assumptions on the plant.