DYNAMIC SEQUENCING OF ROBOT MOVES IN A MANUFACTURING CELL

For manufacturing cells that use a robot for loading and unloading all parts, cell efficiency directly depends on the effective sequencing of robot moves. Since traditional sequential control can only direct the robot to react to service requests on a first-arrival-first-served (FAFS) basis, concurrent models are needed to accommodate multiple events occurring at the same time. In this paper, two dynamic sequencing models are developed with the concurrent modeling capabilities of colored and timed Petri nets. In addition to physical activities in the cell, sequencing decision processes are modeled as information flows by non-physical objects and integrated into a unified Petri net framework for cell control. To reduce robot idle time, anticipated moves can be made to the next service location. Several prospective service requests can also be predicted and evaluated simultaneously within a time window, and the best move sequence with minimum total robot move time can be determined by considering the robot pick-up and drop-off locations of the service requests. The effectiveness of the dynamic sequencing models using the concurrent approach is demonstrated by improved cell performance over the sequential control method.