Design of Forcing Functions for the Identification of Human Control Behavior

Abstract

Human control behavior, as exercised during vehicle or aircraft control, can be identified by applying forcing functions with a specific frequency content to excite the human control system. The presence of these forcing functions may affect human control behavior as well, an effect which is not well-understood and might considerably affect the result of experiments. Common metrics, such as spectral shape and bandwidth, have been shown to be insufficient in predicting the forcing function's effect on human control behavior. This paper investigates the effect of forcing function phase on the control behavior, as well as the effect of the forcing function frequencies at which power is present, within a fixed spectral shape. These effects were tested in a human-in-the-loop compensatory tracking experiment. Our experiment showed that the effects ofthe frequencies were substantial. For this reason, a new metric was proposed to better predict the effects of forcing functions on the human control behavior, and to allow comparison between forcing functions with different spectral shapes and frequency contents. The metric considers the variance of the signal's derivatives. To test the validity of the metric, forcing functions with equal derivative variances were used in a compensatory tracking experiment. The outcome of the experiment reveals that the resulting behavior indeed can be considered equal when the variances are equal, even when the forcing function spectral shapes are considerably different.