Decentralized Enforcement of Linear State Specifications for Augmented Marked Graphs With a Coordinator

In real-world scenarios, automated manufacturing systems (AMSs) are often restricted to various types of additional constraints, such as the maximal number of jobs simultaneously allowed. In order to reduce control cost and improve control flexibility, one may wish to synthesize a supervisor / controller that provably enforces the control specifications in an efficient manner. By modeling AMSs as augmented marked graphs (AMGs), this article develops an innovative approach, i.e., decentralized control with a coordinator, to enforce a class of linear state specifications, i.e., generalized mutual exclusion constraints (GMECs). The global AMGs considered in this article are modeled by several subsystems interconnected via shared resources. Local control laws (which enabled transitions should be disabled or fired at each state) are first computed separately by the subsystems themselves, but the one-sided decisions made by local controllers may lead the whole system to a blocking state, namely, decision deadlock. We present a coordinated control policy to avoid such unexpected situations with the aid of a coordinator. It is shown that with our approach, the global GMECs can be enforced in an efficient way based on some limited local information, without knowing any global one.