Comparing virtual environments for cybersickness using a cumulative optical flow entropy metric

Cybersickness, or feelings of nausea, discomfort or unease, are common in virtual reality experiences with head-mounted displays. With the widespread availability of virtual reality headsets across a wide domain of uses including industry, defence, education and the commercial market, is it critical that virtual environments are developed that minimise cybersickness. Unfortunately, determining whether a virtual reality experience will induce cybersickness is difficult. Typically this requires user studies with a completed, or almost completed, virtual environment. This is time consuming and expensive, both to run participant-based user studies and for any rework to the virtual environment needed due to identified issues. As part of modern iterative development processes it would be useful to pre-evaluate virtual environments for cybersickness before engaging user studies. This paper presents a new approach and metric to compare virtual environments’ susceptibility to induce cybersickness. The approach combines visual optical flow, an entropy metric of complexity and a cumulative time-series measure. Virtual environments with known cybersickness attributes are used to demonstrate the approach. Results indicate that the approach can successfully differentiate between known levels of cybersickness and attributes contributing to cybersickness, such as motion direction and field of view.