StickyPie: A Gaze-Based, Scale-Invariant Marking Menu Optimized for AR/VR

Title of the Paper

StickyPie: A Gaze-Based, Scale-Invariant Marking Menu Optimized for AR/VR

Paper Information

• Subject Area: Interaction design for Augmented Reality (AR)/Virtual Reality (VR), gaze input, and marking menu technology
• Keywords: Gaze input, marking menu, AR/VR, head-mounted display, user interface design, eye tracking, interaction techniques, information transfer rate, experimental design

Research Background and Problem

• Identified Issues or Challenges:

  • Challenges exist in menu selection using eye-tracking technology, including overshooting, incorrect selection, and false activations.
  • Current gaze-based interface designs are not optimized to support a smooth transition from novice to expert operation.
  • Most existing research has not deeply explored how to optimize design parameters such as menu size, hierarchical depth, and granularity for gaze-based menus, nor have they addressed marking menu issues in scenarios requiring high precision.

• Significance:

  • The widespread adoption of AR and VR devices has made gaze input an important hands-free interaction method. Optimizing this interaction method can significantly enhance human-computer interaction efficiency and user experience.

• Research Motivation and Related Work:

  • Marking menus, which support both novice users (via graphical user interfaces) and expert users (via gesture operations), offer potential applications for gaze input.
  • Current gaze input research and technologies (e.g., pEyeWrite, OctoPocus) have limitations in user transition design and error handling.
  • Designing interfaces that adapt to varying levels of granularity and depth under limited eye-tracking accuracy is a critical challenge.

Solution

• Proposed Method or Solution:

  • The authors designed a marking menu technique called "StickyPie," which leverages gaze trajectory prediction (particularly saccade landing points) to enable "scale-invariant" input operations. Its core feature dynamically renders the next menu level based on the saccade landing position, avoiding the issue of multiple menu options being triggered simultaneously.
  • StickyPie integrates algorithms to detect saccadic movements (using velocity thresholds and peak detection) and is optimized for users' gaze behavior.

• Innovations:

  • Supports scale-invariant marking input, eliminating the need for precise control of gaze movement distance.
  • Significantly reduces selection errors caused by overshooting, thereby better supporting the transition from novice to expert mode.
  • Experiments demonstrate that StickyPie outperforms traditional methods (RegularPie) in terms of accuracy and learning efficiency.

• Implementation Steps and Key Techniques:

  1. Experimental Design: Conducted multiple experiments to validate the effects of menu size, hierarchical depth, and granularity on user performance.
  2. Algorithm Implementation: Applied velocity detection algorithms to determine saccade start and landing points, optimizing detection thresholds (30°/s for saccade initiation, 10°/s for landing position).
  3. Error Handling: Introduced mechanisms such as detection time limits (200 ms) and blink filtering rules to reduce errors.
  4. Visual Assistance: Provided visual guidance at menu edges to indicate the correct operation path to users.

Research Outcomes

• Specific Results:

  • StickyPie significantly reduced overshooting errors and improved user learning speed.
  • StickyPie outperformed RegularPie in terms of information transfer rate (ITR), especially in expert usage scenarios.
  • The design of StickyPie supports more complex menu hierarchies and gesture designs, enhancing overall system interaction efficiency.

• Comparative Advantages Over Existing Solutions:

  • Compared to RegularPie, StickyPie avoids issues of multiple option triggers caused by gaze path distance.
  • StickyPie’s menu design offers better scalability and user adaptability while supporting higher interaction precision.

• Experimental or Evaluation Results:

  • Experiment 1: Determined the impact of menu size on novice and expert users, identifying an optimal radial size of 2.5° to 3°.
  • Experiment 2: Explored the effects of menu depth and granularity on input efficiency, finding that granularity should be limited to 6 or fewer when depth exceeds 2.
  • Experiment 3: Compared StickyPie with RegularPie, showing that StickyPie had clear advantages in learning efficiency, information transfer rate, and user preference.

• Limitations and Future Directions:

  • StickyPie requires constraints on the spatial range of menu appearance to ensure visual and operational comfort (recommended range within ±20°).
  • The current solution primarily targets gaze-only input; future work could extend to scenarios combining head-based interactions.
  • Error recovery mechanisms (e.g., undoing incorrect operations) have not been thoroughly explored; future research could incorporate more robust error correction features.

In summary, the StickyPie technique and its associated design guidelines open up new possibilities for gaze-based interaction in AR and VR devices, offering significant theoretical and practical value.