HUMAN SUPERVISORY PERFORMANCE IN FLEXIBLE MANUFACTURING SYSTEMS

The Flexible Manufacturing System (FMS) integrates humans, computers, robots, and computerized machines in order to meet system goals. The objective of the current dissertation is to explore both the optimal allocation of tasks between human supervisor and computers and the optimal number of machines which a supervisor should control in an FMS. As a background to the proposed methodology, the relevant literature pertaining to FMS is reviewed with special emphasis on human-computer communication, multitask performance in FMS in relation to inverted-U theory, attention theory, and human decision capabilities. The hypotheses are proposed that there is an interaction effect between the number of machines and the task allocation level and that there exists an optimal level of number of machines for the operator to monitor. These hypotheses were tested in a statistically balanced experimental design using 30 subjects with three independent variables: number of machines (5 levels--4, 8, 12, 16, or 20 machines), task allocation (2 levels--high or low), and sequence of task presentation (2 levels). Physiological measures (EKG and respiratory rate) and two subjective stress questionnaires were used as indices of the arousal level. The results reveal that the Inverted-U function for response time as a function of machine level was not supported and the allocation level was significant for both performance and subjective stress criteria. These results suggest that in the supervision of FMS the task allocation to the supervisor is critical but that the size of the FMS system (4 to 20 machines), with the allocations studied, has no effect on either stress or performance of the operator in the FMS