Supervisor Localization of Discrete-Event Systems under Partial Observation

Recently we developed supervisor localization, a top-down approach to distributed control of multiagent discrete-event systems. Its essence is the allocation of monolithic (global) control action among the local control strategies of individual agents. In this paper, we extend supervisor localization by considering partial observation, namely not all events are observable. Specifically, we use the recently proposed concept of relative observability (closed under union) to compute the feasible and nonblocking partially-observed monolithic supervisor; then we develop a localization procedure to decompose the monolithic supervisor into a set of local controllers under partial observation. The key concept of the localization procedure is a partially-observed control cover, and in the resulting local controllers only observable events can cause state transitions. The results are illustrated by a Transfer Line example. Further, to deal with large-scale systems, we combine the partially-observed supervisor localization with an efficient heterarchical synthesis procedure; this is applied to study the distributed control of automated guided vehicles serving a manufacturing workcell under partial observation.