VRPursuits
Also known as Pursuits in VR, Smooth pursuit selection
Targets in VR move along their own paths, and the one the user follows with their eyes is selected by matching gaze movement to target movement.
How it works
Each selectable 3D object moves along a distinct trajectory (e.g. circular). Gaze positions are correlated with each target's trajectory using Pearson correlation over 300 ms windows; every 900 ms the mean of the last three coefficients is checked and the best-matching target above a 0.4 threshold is selected, so no calibration accuracy or dwell is needed. In a study (N=26, HTC Vive with Pupil Labs eye tracker) performance was robust to target size and distance but improved with larger trajectories, especially while walking; selection while walking worked but was less accurate. Demonstrated with a PIN-entry ATM and an asteroid-shooting game.
Papers
- Mohamed Khamis, Carl Oechsner, Florian Alt, Andreas Bulling (2018). VRpursuits: Interaction in Virtual Reality Using Smooth Pursuit Eye Movements. Proceedings of the 2018 International Conference on Advanced Visual Interfaces (AVI '18) doi:10.1145/3206505.3206522