Most research on manufacturing automation focuses on the derivation of fully automated control and scheduling techniques. An alternative and more realistic paradigm to "lights out" automation is supervisory control; the design of manufacturing control and scheduling processes that explicitly integrate human decision makers with underlying automation. Supervisory control designs the human-computer interaction in order to augment and to extend the human's role and decision making effectiveness. Neither the goal nor the unintended side-effects of supervisory control are to automate the human decision maker out of the system nor to reduce the human's role to a set of undesirable or ineffective tasks. At the same time, supervisory control ameliorates the inherent limitations of flexibility in automated control. This research offers supervisory control as an alternative to the goal of full automation in manufacturing processes. With supervisory control, the goal is to design into the control process human functions that utilize human skills and that enhance human effectiveness and overall system performance. The research offers a three part approach to explore supervisory control in advanced manufacturing systems: first, the development of an interactive, real-time, object-oriented manufacturing simulator; second, the development of modeling methods to prescribe/describe operators who control and troubleshoot manufacturing processes; and third, to use the operator models and the object-oriented manufacturing simulator to systematically explore, with realistic manufacturing system dynamics, the integration of analytical methods and knowledge-based techniques, effective designs for an on-line decision support system.
