Nonlinear Prediction of Head Movements for Helmet-Mounted Displays

Abstract : In head-and eye-slaved visual systems, lag times in the visual feedback loop are more apparent than they are in conventional fixed display systems. The available technology of digital image generators does not permit lag times to be reduced to the required amount. Therefore, appropriate prediction algorithms have to be developed. Accelerometers were used to measure the step response of the head in three axes of rotation. It could be shown that linear prediction does not provide the necessary accuracy in the simulated position. A further analysis of the recorded data revealed that it is possible to take advantage of the head's latency to improve the prediction. A simple nonlinear prediction algorithm based on acceleration data was successfully implemented in the fiber optic helmet-mounted display. (Author)