Estimating Cognitive Load and Cybersickness of Pilots in VR Simulations via Unobtrusive Physiological Sensors

Abstract

Predicting real-time estimates of cognitive load in pilots assists intelligent flight systems in alleviating high workloads, thereby averting accidents and directly impacting safety in aviation. Virtual Reality (VR) flight simulations provide an immersive stage to evaluate physiological measures and identify their cognitive correlates. In this work, unobtrusive sensors such as eye-tracking, pupillometry, and photoplethysmography (PPG) record physiological data while six participants perform six flying tasks of varying complexity in VR. The extracted feature sets such as pupil diameter change, number of fixations and saccades, and heart rate variability (HRV) are compared to the Pilot Inceptor Workload (PIW) measures, specifically duty cycle and aggressiveness. The PIW, number of saccades, and the self-reported workload measures were significantly affected by the tasks. However, the number of saccades measure demonstrated a significant negative correlation with the PIW’s measures, contradicting prior work. The remaining feature sets, including the pupil diameter change and the number of fixations, display a nearly identical trend to the PIW measure, though no significance was detected.