Computational Complexity for State-Feedback Controllers with Partial Observation

Abstract : We study the computational complexity of several decision and optimization control problems arising in partially observed discrete event systems. These problems are related to the state avoidance problem where one must compute a controller which prevents the system from accessing a set of bad states and which is maximal for a defined criterion, based on inclusion of the set of states remaining reachable after the control. We focus our study on memoryless controllers.