Scheduling and control of Flexible Manufacturing Cells (FMCs) have been of continuing theoretical and practical importance to the operational-research and automatic-control communities, respectively. Petri Nets (PNs) and controlled-automata have been two commonly used techniques for FMC modeling and supervisory controller synthesis. However, although the literature contains a large number of PN-based scheduling methods, there have been no attempts to utilize automata theory for similar endeavors. This thesis is one of the first reported research efforts for using automata theory for simultaneous scheduling and control of FMCs. In order to verify the effectiveness of the proposed methodologies, a MS-Windows based software was developed to simulate the implementation of the methodologies for different FMC configurations. Numerous simulations clearly demonstrated that the proposed methodologies can control FMCs and improve their performances simultaneously.

An automata-based supervisory controller ensures that the plant under supervision will behave deadlock-free (according to given specifications). However, it lacks decision-making ability. The overall research objective of this thesis, thus, is to develop automata-based methodology for synthesizing FMC controllers capable of monitoring, controlling, and making scheduling decisions such that they could control the flow of parts within the workcell, satisfy the deadlock-free requirements, and improve the FMC performance simultaneously. As proposed in this thesis, first, time-augmented automata are used for system modeling purposes. Supervisory-control theory is, then, used as the supervisor-synthesis method. The constructed supervisor, representing deadlock-free FMC behavior, is utilized to search for the optimized workcell behavior. For the above-mentioned three problems an A* based search methodology, a Non Linear Programming (NLP) modeling method, and an on-line beam-search based methodology are utilized, respectively.

Three formal methodologies are proposed to solve three basic control and scheduling problems, respectively. The first problem deals with off-line scheduling for batch production. The second problem deals with continuous production, where parts, one by one, are continuously introduced into the workcell. It is assumed that only the probability of arrival of the incoming part is known in advance. The third problem deals with random-arrival production, where the controller has no a priori information about the arrival of parts and has to evaluate the state of the workcell and determine possible available alternative decisions in an on-line manner.

As proposed in this thesis, first, time-augmented automata are used for system modeling purposes. Supervisory-control theory is, then, used as the supervisor-synthesis method. The constructed supervisor, representing deadlock-free FMC behavior, is utilized to search for the optimized workcell behavior. For the above-mentioned three problems an A* based search methodology, a Non Linear Programming (NLP) modeling method, and an on-line beam-search based methodology are utilized, respectively.

In order to verify the effectiveness of the proposed methodologies, a MS-Windows based software was developed to simulate the implementation of the methodologies for different FMC configurations. Numerous simulations clearly demonstrated that the proposed methodologies can control FMCs and improve their performances simultanously.