Reducing Lag in Virtual Displays Using Multiple Model Adaptive Estimation

摘要

Multiple Model Adaptive Estimation is an effective method for reducing lag invirtual environment displays. Lag in displays (the time from head motion to the appearance of the proper image on the display) is a significant detriment to realism in virtual environments. Increasing the speed of the computers which control the virtual display is not a final answer. No matter how fast the processors work, there will always be demands to do more. Predicting angular head positions (look-angles) can reduce the lag by allowing the computer to calculate the appropriate scene before it is needed on the display. Single predictors cannot adequately cover the dynamic range of head motion. Using a bank of three elemental filters, lag in head orientation can be significantly reduced when compared to systems with no predictor or the single, nonadaptive Kalman filter predictor proposed by Liang. Predictions lead to small overshoots when the angular velocity of the head reverses direction, but the overshoot lasts for only one frame and is not significantly larger than the nonadaptive predictor.