We propose a method that dynamically adjust the dwell time for each hyperlink in a gaze-based web browser using a two step selection policy. We first estimate the probability of each hyperlink of being the intended target of the user by hidden Markov model based on the user's past gaze trajectory. We then assign different dwell times to different hyperlinks, where a hyperlink with a higher probability of being the target is assigned a shorter dwell time.
The graphical user interface is shown below.
To select a hyperlink, the user should:
(1) Look at the 'Select' button for 500ms to trigger it;
(2) Look at the hyperlink s/he wants to select, where the dwell time varies.
(1) A two-stage hidden Markov model is adopted to model the natural gaze trajectory and estimate the probability of each hyperlink being the intended target.
(2) A set of piecewise linear functions with four parameters is adopted to assign dwell time based on the probability.
The demo of this work can be found in https://www.youtube.com/watch?v=N1GGn1O8wlM.
@article{chen2019using,
title={Using variable dwell time to accelerate gaze-based web browsing with two-step selection},
author={Chen, Zhaokang and Shi, Bertram E},
journal={International Journal of Human--Computer Interaction},
volume={35},
number={3},
pages={240--255},
year={2019},
publisher={Taylor \& Francis}
}