Lattice Menu: A Low-Error Gaze-Based Marking Menu Utilizing Target-Assisted Gaze Gestures on a Lattice of Visual Anchors

Title of the Paper

Lattice Menu: A Low-Error Gaze-Based Marking Menu Utilizing Target-Assisted Gaze Gestures on a Lattice of Visual Anchors

Paper Information

• Domain: Human-Computer Interaction - Eye Tracking and Interactive Interface Design
• Keywords: Eye Tracking, Gaze Interaction, Marking Menu, AR/VR, Human-Computer Interaction Design

Research Background and Problem Statement

• Issues and Challenges:

  • Traditional gaze-based marking menus (e.g., boundary-crossing methods) suffer from accuracy issues, particularly facing high error rates.
  • The use of complex multi-level menu structures in the field of gaze tracking is often limited by user biases or inaccuracies in eye movements.
  • The lack of fixed visual targets (e.g., "in-air" gaze gestures in boundary-crossing methods) easily leads to errors.

• Significance:

  • With the development of technologies like AR/VR, eye-tracking technology is becoming increasingly important, especially in hands-free interaction scenarios. Researching a gaze interaction system with low error rates and fast selection can enhance user experience, particularly in multi-level menu structures, by reducing user fatigue.

• Motivation and Related Work:

  • Boundary-crossing methods and other gaze-based interaction techniques, such as EyeWrite and EyeDraw, have attempted to support precise gaze gestures but have not achieved sufficiently low error rates.
  • New methods like StickyPie have improved some issues but still require further optimization, especially in controlling errors during multi-level menu item selection.

Proposed Solution

• Proposed Solution:

  • Lattice Menu: A gaze gesture-based marking menu that improves accuracy and reliability by utilizing target-assisted gaze gestures on a lattice of visual anchors.

• Innovations:

  • By providing a "lattice of visual anchors," users' gaze has clear targets, enabling stable and accurate gaze gestures.
  • Developed an interaction scheme that significantly reduces error rates (approximately 1% error rate).
  • Designed a progressive unfolding effect for visual anchors to minimize inconvenience caused by visual distractions.

• Implementation Steps and Key Techniques:

  • Optimized the visual design parameters of the menu through experiments (e.g., anchor size, item selection zone range, menu dimensions).
  • Implemented the Lattice Menu in a VR environment using FOVE eye-tracking equipment and designed methods to measure task completion and error rates, comparing performance under different conditions.
  • Conducted multiple user studies, including comparisons of Lattice Menu with traditional gaze menus (e.g., BorderPie) in terms of error rates and selection time.

Research Outcomes

• Specific Results:

  • Error Rate: The error rate of Lattice Menu in multi-level structures was significantly lower than traditional methods, e.g., only 1% in a 4×4×4 menu structure.
  • Selection Time: In expert user scenarios, menu selection time ranged from 1.3 to 1.6 seconds, faster than other methods.
  • User Experience Improvement: Post-study interviews revealed that all participants preferred the Lattice Menu, citing that visual targets improved accuracy and reduced fatigue.

• Experimental and Evaluation Results:

  • Compared to traditional methods without visual anchors (e.g., BorderPie), the Lattice Menu reduced error rates by a factor of 5 in expert use cases.
  • Results showed that the Lattice Menu improved menu selection speed by approximately 1.3 times and significantly reduced error rates.
  • Users' gaze landing points were evenly distributed near the visual anchors, further validating the effectiveness of visual targets.

• Limitations and Future Directions:

  • Limitations:
    • The current design was primarily tested in VR headset environments with fixed backgrounds, without considering the impact of complex dynamic backgrounds on user experience.
    • Further exploration is needed to optimize operations for "cancel" or "undo" options.
  • Future Directions:
    • Explore the differences between head-fixed menus and eye-only control to support broader applications.
    • Develop customizable Lattice Menus for different users, such as adaptive visual anchor sizes or precision optimization.
    • Test and adjust the system in various real-world application scenarios (e.g., dynamic AR/VR environments or smart TV interactions).