Human-in-the-Loop Spacecraft Maneuvers: Investigating Human Factors Through VR and Machine Learning

On-orbit manual attitude control of manned spacecraft is accomplished using external visual references and some method of three axis attitude control. All past, present, and developmental spacecraft feature the capability to manually control attitude for de-orbit. We used a Unity Virtual Reality (VR) spacecraft simulator to accurately achieve de-orbit attitude while in a 400 km circular orbit. We compared three different devices to control attitude - a 3-DoF Flight stick, physical buttons and virtual buttons with haptic feedback. While the flight stick was fastest in terms of deorbiting, physical buttons were most economical in terms of fuel consumption. Comparison was based on time taken, fuel consumed, cognitive workload assessment and user preference. We used ocular parameters, EEG, NASA TLX and IBM SUS to quantify our results. We also compared a set of machine learning models to predict self reported cognitive load from EEG and pupil dilation.