Operator-Centered Adaptive Compensation of Continuous Manual Control Systems

Abstract

The subject of this paper is an adaptive compensator, which has the function of minimizing the compensation required of the human operator in a continuous single-dimensional manual control system. A random input and a compensatory display are assumed. Use of this compensator does not require detailed knowledge of system dynamics since its adaptive action is based solely upon the operator's control actions. A quasi-linear operator model and the parameter-tracking model method of measurement are used to provide the required on-line mathematical description of the operator's control actions. Design of the compensator is based upon past results, which show the human operator to be an adaptive controller who introduces compensation, via his control actions, similar to that which would be chosen by a servo-engineer in minimizing the square of the system error. The compensator's function results from the established fact that an operator can control a system most accurately when the least compensation is required of him. Experimental results demonstrate that the compensator reduces system error while simultaneously simplifying the operator's task. They also establish some of the limitations of the compensation scheme as currently implemented.