Visual Alignment Accuracy in Head Mounted Optical See-Through AR Displays: Distribution of Head Orientation Noise

The mitigation of registration errors is a central challenge for improving the usability of Augmented Reality systems. While the technical achievements within tracking and display technology continue to improve the conditions for good registration, little research is directed towards understanding the user's visual alignment performance during the calibration process. This paper reports 12 standing subjects' visual alignment performance using an optical see-through head mounted display for viewing directions varied in azimuth (0°, ±30°, ±60°) and elevation (0°, ±10°). Although viewing direction has a statistically significant effect on the shape of the distribution, the effect is small and negligible for practical purposes and can be approximated to a circular distribution with a standard deviation of 0.2° for all viewing directions studied in this paper. In addition to quantifying head aiming accuracy with a head fixed cursor and illustrating the deteriorating accuracy of boresight calibration with increasing viewing direction extremity, the results are applicable for filter design determining the onset and end of head rotation.