EEG-Based Assessment of Cybersickness in a VR Environment and Adjusting Stereoscopic Parameters According to Level of Sickness to Present a Comfortable Vision

Abstract

Virtual reality (VR) is an increasingly widespread technology that provides a more realistic and fully immersive experience by using head-mounted displays (HMDs). However, this medium comes with some side effects. Users immersed in a virtual environment (VE) experience motion sickness like discomfort, which is named visually induced motion sickness(VIMS) or, more commonly, cybersickness. In an effort to overcome cybersickness experienced with stereoscopic displays, we propose a novel real-time system to detect cybersickness from the incoming electroencephalogram (EEG) feedback and to mitigate it by updating the cue parameters as per the feedback from the proposed model. The VE used in the study was generated procedurally by tuning levels of 3 different types of cues (navigation speed, scene complexity, and stereoscopic rendering parameters) to induce cybersickness in a varying range of severity. In the first phase of the study, we trained a two-stage shallow convolutional neural network with the EEG data collected from the users while immersed in the VE. The proposed two-stage model was utilized to detect cybersickness and to classify factors causing cybersickness, respectively. The performance of the cybersickness detection model reached an overall accuracy of 76.26%, while the factor type classification model achieved 81.01% overall accuracy. To assess the performance of the proposed cybersickness detection and mitigation system, an experiment consisting of two control sessions, and one models-in-the-loop session (MIL) was conducted in the second phase of the study with a different user sample. The differences in the Simulator Sickness Questionnaire (SSQ) responses collected before and after each session, and the time-dependent changes in the cue parameters showed that the participants felt less cybersickness during the MIL session in which the proposed cybersickness detection and mitigation system (CDMS) was utilized.