Abstract by Molly Barrott

Personal Infomation

Presenter's Name

Molly Barrott


Adam Riley

Degree Level



Adam Riley

Abstract Infomation


Computer Science

Faculty Advisor

Parris Egbert, Seth Holladay


Avatar Location Prediction in Virtual Reality


While virtual reality (vr) has made strides in recent years, most consumer-grade, vr headsets have limited user tracking ability. Consequently, almost all consumer vr experiences are lacking an avatar of the user, leading to a lessened sense of immersion in the vr environment. The purpose of our research is to create a full body, avatar that achieves realistic movements with the limited data we have available. The movements must be generated at 90 FPS to avoid causing motion sickness for the user. We are using knowledge about human kinematics, motion capture data and machine learning to predict where our user’s lower body should be given the location of his head, hands and hip. We then use this information a trained machine model to directly embed a prediction function into our vr experience that will do real-time estimation of what the position and movement of the avatar should be.