Bio-physiological-signals-based VR cybersickness detection

With the gradual maturity of virtual reality (VR) technology in recent years, VR industry is a trend rapid growth, providing new possibilities for content design. Although has been able to provide users with excellent immersive experience, side effects that affect user experience still exist, especially cybersickness. It would cause extreme physical discomfort and discontinuation use. Many researchers have tried find inducement cybersickness detect limit occurrence this symptom, but most current detection analysis methods rely on subjective questionnaires collect users’ posterior states, such as dizziness, nausea, cold sweats, disorientation, eyestrain so on. There no mature real-time system developers evaluate susceptibility their products far, which hindered adoption some extent. The purpose study implement monitoring using physiological sensors measure data quantify influence factors through deep learning model. Besides, we developed experimental platform passive navigation task induce During experiment, train LSTM Attention neural network model, collected user’s signals, including skin electrical activity (EDA) electrocardiogram (ECG), well position bone rotation avatar. model can level during experience. And verified by fivefold cross-validation average accuracy 96.85% was achieved classification level, showing great performance compared other relevant studies. results show feasibility accurate proposed. Also reference improve