Identifying Early Predictors of Learning in VR-based Drone Training

Abstract

The use of small unmanned aerial systems, or drones, has grown rapidly in recent years, but the FAA has no formal requirement for hands-on training. Physical drone training has major limitations, and virtual reality (VR) offers a promising alternative. This study sought to identify early performance markers of learning in VR-based drone training using a physical controller. 14 participants completed a customized VR-based drone training curriculum, while performance metrics, perceptions of workload, and physiological data were collected. Participants were clustered into high and low performers based on a final evaluation task, and separate analyses of variance were conducted to test performance differences between the two groups over time in each training level. We found significant differences in the performance metrics and subjective workload of low and high performers throughout training, suggesting that performance can be predicted early in training and opening the door to future adaptive training systems.