Robust Supervisory control policy for automated manufacturing systems with a single unreliable resource

Over the past two decades, the development of supervisory controllers that guarantee deadlock-free operation for automated manufacturing systems (AMSs) has been an active area of research. Most work to date assumes that the system resources are reliable. This paper focuses on the robust supervisory control problem of AMSs with a single unreliable resource. Our objective is to develop a robust supervisory control policy under which the system can continue producing in the face of the unreliable resource's failure or recovery. To do so, we integrate an optimal deadlock avoidance policy based on a Petri net with a modified Banker's Algorithm and present a novel robust supervisory control policy. It is proven to be of polynomial complexity and more permissive than two existing policies. Also, experimental results on a set of AMSs generated randomly indicate its superiority over all other existing policies.