EyeGuide & EyeConGuide: Gaze-based Visual Guides to Improve 3D Sketching Systems

Document Title

EyeGuide & EyeConGuide: Gaze-based Visual Guides to Improve 3D Sketching Systems

Document Information

• Topic Area: Utilizing gaze tracking technology and interactive guidance to enhance 3D sketching systems in virtual reality (VR)
• Keywords: 3D sketching, gaze tracking, virtual reality, user interface, multimodal interaction, shape accuracy, user study

Research Background and Issues

• Problems and Challenges:

  • In current VR 3D sketching systems, users struggle to accurately position drawing trajectories in three-dimensional space. This stems from high perceptual-motor and cognitive demands, depth perception issues with stereoscopic displays, and the lack of physical support.
  • Existing tools (e.g., GravitySketch and Google Tiltbrush) provide assistance through manually activated 3D grids, but users must switch between control operations and drawing execution.
  • Previous studies have found that adaptive visual guides (e.g., adaptive grids based on handheld controllers) can improve accuracy, but they fail to predict users' future drawing intentions.

• Research Significance:

  • As VR increasingly supports gaze tracking technology, this technology is considered a novel approach to dynamically adjust drawing assistance tools to align with users' attention.
  • Such gaze-based technologies, combined with handheld control techniques, could offer improved human-computer interaction experiences and enhance drawing accuracy and efficiency.

• Research Motivation:

  • Integrating eye gaze movements with manual drawing operations to provide instant and relevant visual assistance without increasing user operational burden.
  • Exploring new gaze-attention-based assistance tool designs aimed at reducing information overload and interference while delivering useful guidance.

Solution

• Proposed Technologies:

  • EyeGuide: A fully gaze-tracking-based drawing assistance method that dynamically displays grid segments in the direction of the user's gaze, with the grid appearing approximately 75 cm away from the user.
  • EyeConGuide: A method that combines gaze tracking and controller position to dynamically adjust the grid segment's location and range, particularly enhancing coverage of existing trajectories and target areas during drawing.

• Innovations:

  • Introducing gaze tracking as the core mechanism for 3D sketching assistance tools.
  • Multimodal interaction techniques (combining gaze and controller) to achieve novel designs in the logic and visual characteristics of dynamic grid appearance.
  • Implementing smooth fade-in and fade-out grid presentation to reduce users' visual burden.

• Implementation Steps and Key Technologies:

  1. Using HTC VIVE Pro Eye VR equipment combined with Tobii gaze tracking technology to extract real-time gaze vectors.
  2. Developing specific display logic and rendering techniques for grid segments, dynamically adjusting display based on gaze position, controller position, and drawn trajectory information.
  3. Validating technological performance through two user studies: basic geometric shapes and complex 3D shape drawing tasks.

Research Outcomes

• Main Results:

  • Accuracy Improvement:
    • Compared to no guidance or traditional handheld grid (SG-Line), gaze-based guidance methods (EyeGuide and EyeConGuide) significantly improved drawing linearity, shape similarity, and accuracy.
    • Users demonstrated better drawing performance without increasing task completion time.
  • User Experience Enhancement:
    • EyeConGuide received high System Usability Scale (SUS) scores (graded "B").
    • Participants generally preferred gaze-assisted methods, particularly EyeConGuide, considering it easier to use and helpful for faster task completion.
  • Impact of Spatial Ability:
    • High spatial ability (HSA) users exhibited more precise drawing, especially under no guidance or EyeGuide conditions.
    • EyeGuide and EyeConGuide significantly improved accuracy for low spatial ability (LSA) users.

• Experimental Data:

  • Experiment 1 (Basic Shapes):
    • EyeConGuide performed best in line straightness and shape deviation, while EyeGuide excelled under specific conditions.
    • 35% of users preferred EyeConGuide, while 17.8% favored EyeGuide.
  • Experiment 2 (Complex Shapes):
    • Both EyeGuide and EyeConGuide outperformed SG-Line and no guidance conditions.
    • EyeConGuide resolved significant shape matching errors, with any type of visual guidance outperforming no guidance.

• Limitations and Future Directions:

  • The participant sample size was relatively small (16 per experiment); future studies should expand the sample scale.
  • The dynamic characteristics of visual guidance (e.g., fade-in/fade-out modes) and their specific impact on user behavior and experience have not been fully explored.
  • Currently validated only in VR environments; future applications could extend to other 3D sketching devices and scenarios, such as desktop 3D sketching tools.