Target selection by gaze pointing and manual confirmation: performance improved by locking the gaze cursor

Abstract

Eye movement-based human-computer interactions are emerging in diverse scenarios. When selecting targets on a user interface, the method of combining fast gaze pointing with reliable manual action is becoming increasingly popular. However, this method suffers from noise in gaze pointing caused by eye jitters and users’ habitual early move-away of gaze before manual actions. Here we propose a novel solution to mitigate these problems by locking the gaze cursor at the target for imminent manual selection. We compared this gaze-lock cursor with a conventional gaze cursor in a typing task with varying key sizes and key gaps. Results show that typing performance was significantly better with larger key size and gap. More importantly, the gaze-lock cursor significantly increased speed and decreased errors when compared to a conventional gaze cursor. Our findings demonstrate that the gaze-lock cursor is a promising tool for gaze interactions involving frequent target selections.

Practitioner summary: Target selection by gaze pointing and manual confirmation suffers from eye jitters and users’ habitual early move-away of gaze before manual actions. The performance of this method can be improved by applying the gaze-lock cursor we proposed, increasing target size or increasing the target gap.

Abbreviations: WTC: warping to target center; ALCM: automatic lock of cursor movement; LCD: liquid crystal display; EWMA: exponential weighted moving average; ER: error rate; ET: execution time; ED: edit distance; CV: coefficients of variation; ANOVA: analysis of variance; GUIs: general user interfaces

Keywords:

• Eye tracking
• human-computer interaction
• gaze cursor
• eye jitters

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The study was supported by the National Key Research and Development Programme of China [No. 2017YFB1300305], Foundation of National Key Laboratory of Human Factors Engineering [No. 6142222180305, SYFD180051802], and National Natural Science Foundation of China [No. 31622029, 31671168, 61533001]. We thank Xinyong Zhang, Erez James Cohen and two anonymous reviewers for their constructive suggestions for the improvement of the manuscript.