A Calibration Interface for 3D Gaze Depth Disambiguation in Virtual Environments

In Augmented and Virtual Reality, accurate eye tracking is a requirement for many applications. Though state-of-the art algorithms have enabled sub-degree accuracy for line-of-sight tracking, one remaining problem is that depth tracking, i.e. calculation of the gaze intersection at various depths, is still inaccurate. In this paper, we propose a 3D calibration method that accounts for gaze depth in addition to line-of-sight. By taking advantage of 3D calibration points and modeling the relationship between gaze inaccuracy and depth, we show that we can improve depth calculations and better determine the 3D position of gaze intersections in virtual environments.