A KNOWLEDGE-BASED SYSTEM TO MODEL HUMAN SUPERVISORY CONTROL IN DYNAMIC PLANNING

摘要

With the increases in the levels of automation and computerization, supervisory control systems are becoming increasingly common in commercial and military applications. A supervisory control system consists of one or more human operators interacting with highly automated components such as those seen in satellite ground control, flexible manufacturing systems, or nuclear power plants. Humans typically perform cognitively intense tasks such as monitoring, planning, real-time control, and troubleshooting, and are ultimately responsible for the safe and efficient operation of the overall system. Although developments on supervisory control have led to useful applications in interface design and automation, there is a scarcity of research that empirically evaluates human decision making in supervisory control through emulation of task performance using knowledge-based systems. In the context of dynamic planning involving simulated search and rescue missions using ground based autonomous robots and uninhabited aerial vehicles, we developed a knowledge-based system that mimics supervisory control performance. This paper describes the application domain, the details of the simulation model, and the implementation and assessment of a knowledge-based system that mimics human supervisory control performance.