Supervisory Control, Mental Models and Decision Aids

Abstract

This paper poses a framework for considering human supervisory control of semi-automatic systems, a framework which extends earlier definitions and paradigms. It analyses supervisory control into specific human functions and gives examples of research that have been done and/or are needed with respect to each of these functions. For each such function it is argued that the human supervisory operator necessarily has a corresponding mental model, and potentially can have a computer-based decision aid. The relation of the proposed framework to the canonical modern control paradigm is also discussed, as are the reasonable limitations of our ability to model such a complex human machine interaction which itself exercises a high degree of free choice.