Identification of Finite State Models of A Human Operator.

Abstract : The problems of modelling a class of manned systems in which the operator or operators have available only a finite number of decision alternatives which they can use to control the system configuration and mode of operation over time are the focus of this report. An abstract system theoretic structure suitable for representing such discrete control systems is developed and the structure is used to organize the analysis of data obtained from a man-in-the-loop simulation of an AAA system. The structure used to define discrete control is a hierarchical/heterarchical network of finite state systems. The nodes in this network represent system components, task and activities. Several levels of abstraction are used which means that both macroscopic and microscopic descriptions are possible. The structure captures the certain aspects of coordination and the flow of decision making activity through the system. The feasibility of discrete control modelling is demonstrated. The structural aspects, particularly knowledge representation and the identification of key decision points, seem quite powerful. The statistical and data analysis procedures work successfully but need further refinement if model construction is to maximally efficient.