HUMAN SUPERVISORY CONTROL OF A FLEXIBLE MANUFACTURING SYSTEM: AN EXPLORATORY INVESTIGATION

Flexible Manufacturing Systems (FMSs) have provided a means for efficient and reliable automation of batch manufacturing. The purpose of the proposed study was to explore human supervisory capabilities in controlling a FMS in order to plan the future job design of supervisory controllers of these systems. Two theoretical frameworks were incorporated into the proposed study in order to evaluate human control capabilities. One of these was reflected in task conditions characterized by hypothetically distinct dimensions of the mental workload construct. The second framework consisted of a two-stage sequential decision making model that had intrinsic to it the imposition on the human of a relatively well-defined locus of control. In addition, the usefulness of decision support aimed at enhancing human pattern-recognition capabilities was explored by making a graphical display of the FMS available under certain task conditions. An experiment was performed employing a real-time interactive simulation model of a FMS. Through a set of menus, 12 subjects were able to obtain information from and execute control actions to the system. Subjects performed the supervisory task for seven one-hour sessions, and were instructed to optimize a set of criteria related to FMS performance. Results indicated that the human appeared better at controlling the FMS under task conditions characterized by an externally induced workload (i.e., increased activity within the FMS) compared to conditions where the workload was more internally induced (i.e., brought about by requiring the human to optimize an additional criterion). Under conditions where breakdowns occurred in the human\u27s interface to the FMS, performance improved when the human adopted a conservative response strategy characterized by lower levels in the locus of information acquisition and control. Additionally it was found that the particular graphics that was made available generally did not prove very effective as a decision-support device. Evidence demonstrated that when graphical information was made available in addition to the alphanumeric information that was present in the menus, the relative usefulness of each type of information depended on whether well-defined or more generalized criteria were being optimized