Browse Teach Curate

Vision-Only Cybersickness Prediction via Cross-Modal Alignment in Virtual Reality

A deployable framework that predicts an individual's VR cybersickness from consumer-headset signals. Students learn how modality-specific graph neural networks with a difference-attention module encode head-motion, eye-tracking, and physiological time series, and how a cross-modal alignment objective trains a video encoder to match those sensor representations so that, at inference, cybersickness can be predicted personalized from video alone at real-time latency.

KITT Text

Towards Consumer-Grade Cybersickness Prediction: Multi-Model Alignment for Real-Time Vision-Only

Cybersickness is a visually induced form of motion sickness (dizziness, nausea, discomfort) that arises in immersive virtual reality and limits its adoption. Prior high-accuracy predictors depend on …