Study of Cybersickness Prediction in Real Time Using Eye Tracking Data

Abstract

Cybersickness seriously degrades users' experiences of virtual real-ity (VR). The level of cybersickness is commonly gauged through a simulator sickness questionnaire (SSQ) administered after the expe-rience. However, for observing the user's health and evaluating the VR content/device, measuring the level of cybersickness in real time is essential. In this study, we examined the relationship between eye tracking data and sickness level, then predicted the sickness level using machine learning methods. Some characteristics of eye related indices significantly differed between the sickness and non-sickness groups. The machine learning methods predicted cybersickness in real time with an accuracy of approximately 70%.