Pilot Mental Workload Calibration

Open Access

Abstract

Predicting high pilot mental workload is important to the U.S. Air Force because lives and aircraft can be lost when errors are made during periods of mental overload and task saturation. Current research efforts use psychophysiological measures such as electroencephalography, cardiac, ocular, and respiration measures in an attempt to identify and predict mental workload levels. The primary focus of this effort is the development of a calibration scheme that allows a small subset of salient psychophysiological features developed using actual flight data for one pilot on a given day to accurately classify pilot mental workload for a separate pilot on a different day. To accomplish this objective, the signal-to-noise ratio feature screening method is employed to determine the usefulness of 151 psychophysiological features in feed-forward artificial neural networks. Factor analysis identifies patterns in features that vary with changes in workload level. Methodologies for workload level modification and data calibration are presented and tested. Our results indicate the calibration scheme can increase classification accuracy (CA) over 55%, decrease CA variance by 88%, and decrease by 88% the number of features to process than previous classification methods and classifiers.