Reduced Complexity Control Systems

This paper discusses methods for the analysis and design of systems controlled, at least in part, by finite state controllers, such as might be found in adaptive or highly autonomous robotic systems. We focus on the role that feedback can play in simplifying the characterization of trajectories and, in particular, the extent to which elementary feedback rules based on finite state automata can be used to reduce the complexity of both the controller and the analysis. Finally, we introduce a new control paradigm based on randomized finite state controllers and present an analysis of a class of such systems.