Compensation for end to end delays in a VR system

In virtual reality applications, time delay is one of the essential factors which need compensation. We measured the end to end time delay of our system and developed a compensation technique using a head motion model and a predictive Kalman filter. Compared to a previously developed predictive filter with a model free approach, our filter reduces the error, by 20 to 30%. Due to recent advances in PC technology, it was possible to implement the entire system (including the compensation) on a single PC. Our system consists of a PC, head mounted display, and a 6 degrees of freedom magnetic tracking sensor.