A formal model for analysis and control of human-involved computer integrated manufacuring systems

by 1971- Shin, Dongmin

iii Two separate and very distinct approaches to research related to factory control have been pursued – automatic control and man-machine interactions. In spite of the significant work conducted in both areas, very limited crosscutting research has been conducted in this important integration area. It is still true that human integration into highly automated manufacturing environs can be cumbersome and result in system anomalies (crashes and illogical transitions). Research in this critical integration area has likely been limited due to the nature and tools used in these two areas. The focus of the proposed research is to develop formal models and controllers that utilize previous work in both of these critical areas in order to create a comprehensive framework for controller modeling and development containing both automatic and human components. The proposal offers a vision of how a human can fit within a computer-control framework so that the system can take full advantage of tasks and activities that humans do well, and those that computers perform more effectively. The specific classes of human activities addressed herein include: 1) dynamic decision making and 2) physical material handling. The environment that is modeled is one that contains CNC machines, robots, automatic storage systems, and humans. To address the central issue of the proposed work, an automata-based system control model is used to control flexible automated equipment on the shop-floor. Hooks for integrating the human into the controller are modeled by creating a descriptive model using finite state methods where a set of possible human activities is developed. An operator function model characterizing the behavior of human in the various shop-floor iv roles is developed so that a logical partition of activities can be created. Transition methods and software tools that allow the human and computer system to interact seamlessly are developed for a general class of automata and operator function models. It is critical to identify how parts are handled in a system so that a system controller for the human-automation cooperative system can consider possible situations caused by a human operator. For this reason, the impact of human activities on the control complexity from a part states’ view is addressed. The number of possible part flows at a certain state is used to measure the control complexity. Finally, a structure specification of a human-automation interaction and a formal approach to describe the operational features of the structure are presented from a control-theoretic perspective.