GazeBlend: Exploring Paired Gaze-Based Input Techniques for Navigation and Selection Tasks on Mobile Devices

TL;DR

Facing the limitations of single gaze-input techniques for mobile interaction, this paper investigates pairing three gaze-based inputs (Dwell Time, Pursuits, Gaze Gestures) for navigation and selection tasks in seated and walking contexts. Through a 24-participant within-subject study testing six technique-pair configurations, the authors quantify performance (time, completion, errors) and workload (NASA-TLX) and collect qualitative impressions. Key findings show that gestures for navigation paired with dwell time or Pursuits substantially improve performance in walking, with DwellGestures emerging as the most preferred pairing, and that mis-activations are reduced when pairing is used. The work offers design implications and future directions for robust, context-adaptive gaze-enabled mobile interfaces.

Abstract

The potential of gaze for hands-free mobile interaction is increasingly evident. While each gaze input technique presents distinct advantages and limitations, a combination can amplify strengths and mitigate challenges. We report on the results of a user study (N=24), in which we compared the usability and performance of pairing three popular gaze input techniques: Dwell Time, Pursuits, and Gaze Gestures, for navigation and selection tasks while sitting and walking. Results show that pairing gestures for navigation with either Dwell time or Pursuits for selection improves task completion time and rate compared to using either individually. We discuss the implications of pairing gaze input techniques, such as how Pursuits may negatively impact other techniques, likely due to the visual clutter it adds, how integrating gestures for navigation reduces the chances of unintentional selections, and the impact of motor activity on performance. Our findings provide insights for effective gaze-enabled interfaces.

Figures (10)

• Figure 1: This work introduces and evaluates pairing gaze-based techniques: Dwell time, Pursuits, and Gestures for navigation and selection tasks on mobile devices, evaluating user perceptions within a single interface. We tested our techniques under two motor activities: while seated (a) and walking (d). (b) shows a participant navigating soundtracks via Gaze Gestures and selecting via Pursuits (arrows for illustration), and finally (c) shows the study location.
• Figure 2: A step-by-step showing how participants perform the task for the DwellPursuits condition. In the DwellPursuits condition, Dwell time handled selection and Pursuits handled navigation: (1) navigate from page 1 (left) to 2 (middle) via Pursuits in the home screen, (2) select the music player with Dwell time (middle), (3) navigate four screens to the target soundtrack via Pursuits (right), and (4) play it with Dwell time (a selection step). Navigation buttons remain on-screen for consistency, though unused with Gestures.
• Figure 3: This figure represents the mean task completion time in seconds. No significant differences were observed between the techniques during seated motor activity. While walking revealed significant differences between several pairs, DwellGestures and PursuitsGestures showed faster completion time than PursuitsPursuits. The error bars represent the standard deviation.
• Figure 4: The figure illustrates the task completion rate by input techniques and motor activities. DwellGestures and PursuitsGestures resulted in a significantly higher completion rate compared to PursuitsPursuits when walking.
• Figure 5: Sitting and walking task error rates. For tasks performed while seated, DwellGestures demonstrated significantly higher accuracy than DwellPursuits, PursuitsDwell, PursuitsGestures, and PursuitsPursuits. In contrast, during walking, DwellDwell, DwellGestures, and PursuitsGestures significantly outperformed PursuitsPursuits in terms of accuracy. The error bars represent the standard deviation.