Integrating a Computational Model of Optical Flow into the Cybersickness Dose Value Prediction Model

Abstract

This paper describes a computational model of optical flow and demonstrates how it can be used to estimate the navigation velocity of the user's viewpoint during a virtual reality (VR) simulation. The model takes as input a VR simulation video sequence and outputs the estimated average navigation velocity. When this estimated velocity is integrated into the published cybersickness dose value model (CSDV: So, 1999), the severity level of the cybersickness associated with the VR simulation can be predicted. Results of two simulation studies indicated that the average navigation velocities in lateral and vertical axes can be estimated to within 10% error. The simulations used a virtual environment (VE) with a known single depth. Discussion of an extension of the computational model to accommodate VEs with unknown multiple depths is included. The reported model is consistent with the biological neuro-pathway of the perception of visual motion velocities in humans. The benefits associated with this consistency are discussed.