Automatic generation of control software in computer-integrated manufacturing

The rate of acceptance of Computer Integrated Manufacturing and Flexible Manufacturing Systems has been in principle, enthusiastic. However, the rate of implementation of these expensive systems has been disappointing. The bottleneck process in the creation of a CIM facility is the development of software. This thesis presents a theoretical framework that has been used to successfully generate a hierarchy of control software for a full-scale manufacturing facility. A state space model of manufacturing cells is presented which establishes a precise graph model of cooperative actions between material handling robots, processing machines, autonomous guided vehicles, decouplers, conveyors, and automated storage and retrieval systems. The state space model is automatically constructed from a simple input specification, and is shown to be equivalent to a pushdown automaton. This equivalence is of fundamental importance since, by implication, context-free grammars must exist which are capable of controlling manufacturing systems as a byproduct of semantic processing associated with recognizing commands. These grammars are termed context-free control grammars (CFCG). Automatic procedures for constructing CFCG from the state space model are developed, and a systematic method or attaching semantic actions to productions in the control grammars is presented, permitting the automatic generation of finite state control parsers by existing parsing function generators. An efficient representation of process plans is developed for use by area and cell control modules, providing a bridge between control and scheduling activities. The architecture of CIMGEN, an automatic cell and workstation control software generator for computer-integrated manufacturing systems, is presented. The thesis concludes with a discussion of implementation experience, using CIMGEN to produce the distributed controllers for a three-cell ten-device manufacturing system.