Structural Model of the Adaptive Human Pilot

Abstract

A compensatory tracking model of the human pilot is offered, which attempts to provide a more realistic representation of the human's signal processing structure than that which is exhibited by pilot models currently in use. Two features of the model distinguish it from other representations of the human pilot. First, proprioceptive information from the control stick or manipulator constitutes one of the major feedback paths in the model, providing feedback of vehicle output rate due to control activity. Implicit in this feedback loop is a model of the vehicle dynamics which is valid in and beyond the region of crossover. Second, error rate information is continuously derived and independently but intermittently controlled. An output-injected remnant model is offered and qualitatively justified on the basis of providing a measure of the effect of inaccuracies such as time variations in the pilot's internal model of the controlled-element dynamics. The data from experimental tracking tasks involving five different controlled-element dynamics and one nonideal viewing condition were matched with model-generated describing functions and remnant power spectral densities.