Active diagnosis with observable quiescence

Active diagnosis of a discrete-event system consists in controlling the system such that faults can be detected. Here we extend the framework of active diagnosis by introducing modalities for actions and states and a new capability for the controller, namely observing that the system is quiescent. We design a game-based construction for both the decision and the synthesis problems that is computationally optimal. Furthermore we prove that the size and the delay provided by the active diagnoser (when it exists) are almost optimal.