A visualized human-computer interactive approach to job shop scheduling

Abstract

In this study, we propose a human-computer interactive scheduling approach in which humans are supported by a graphic interactive scheduling system. The scheduling system provides humans with a computerized Gantt chart that contains many aiding features under static job shop environments. We compared the performance of humans in the graphic interactive scheduling system with that of commonly used dispatching rules. The results showed that human schedulers were superior to dispatching rules in a multiple objective situation. Next, the optimization behaviour of computer-aided human schedulers in scheduling tasks was observed in an experiment. The experiment investigated the effect of providing a computer-generated initial solution versus having the human subjects generate their own. The results showed that in complex problems the subjects performed significantly better when working with initial solutions that were generated by themselves. This implies that computer aiding that is incoherent to human problem-solving strategies may be less effective than commonly expected. The overall results suggest that the human-computer interactive scheduling using a graphic interface is effective and promising provided that the graphic information is well designed to match human strategies.