An Implicit Method for Dynamic Task Allocation Between Man and Computer in Supervision Posts of Automated Processes

Abstract

The availability of the large production processes depends essentially on the supervision operator's abilities in fault detection, diagnosis and trouble-shooting. One possible solution to insure the process reliability is a man-computer cooperation by means of a dynamic allocation of the decision tasks between them.This paper presents an implicit method for dynamic allocation based on the optimal control theory, with a view to optimizing a performance criterion accounting for production process results and human workload as well. Workload is estimated on line by an operator's observer model which receives the same information as the man, plus the actions performed by him. The model deduces the task demands and the human occupation rate and then computes the sampled workload in multitask situations.A feasibility study realized on an experimental platform shows the validity of the workload assessment method and the existence of an optimal allocation of the tasks between man and computer.