On-line optimal compositional controller synthesis for AGV by unfolding

:This paper deals with the on-line design of an optimal supervisor to coordinate an automated guided vehicle (AGV) fleet based on the γ-unfolding. The γ-unfolding results in an automaton, which permits both to represent the behavior of each vehicle on the considered circuit, and to determine the cost of each admissible sequence directly, without the use of any optimization algorithm. It is based on the definition of energy criteria for the AGVs. We propose an on-line compositional procedure to calculate the optimal supervisor from the γ-unfolded automata modeling the different vehicles. The resulting supervisor is the least restrictive, ensures controllability and nonblocking, and minimizes the energetic cost of the AGV travels.