Exercise Speed Analysis Based on Head Mounted Display Orientation

Virtual Reality (VR) exergames are a useful tool for improving users' exercise levels. However, they have some associated equipment requirements which are an impediment to adoption outside of specialised locations. In our research we aim to make VR exergames more accessible by removing such equipment requirements by determining users' exercise intensity directly from affordable and widely available VR equipment such as a smartphone-based Head Mounted Displays (HMD). We present an approach for determining a user's cycling speed on a stationary exercycle by observing changes to the orientation of the user's view in the VR environment. The approach is computationally cheap and does not require training data sets or server access such as deep learning-based solutions. We evaluate this approach by comparison with speed data captured during exergame play, and through direct use in an exergame. Our results indicate that it has generally good accuracy under most scenarios. In particular, estimated periods of acceleration and deceleration closely match measurements directly collected from the exercycle, but accuracy substantially decreases when large scale variation in the user's view direction is introduced.