FocusFlow: 3D Gaze-Depth Interaction in Virtual Reality Leveraging Active Visual Depth Manipulation

Document Title

FocusFlow: 3D Gaze-Depth Interaction in Virtual Reality Leveraging Active Visual Depth Manipulation

Document Information

• Topic Area: Gaze depth and interaction design in virtual reality
• Keywords: gaze interaction, visual depth, virtual reality, 3D user interface, eye tracking

Research Background and Problem

• Research Problems and Challenges:

  • Current gaze-based virtual reality (VR) interactions primarily utilize gaze direction information, neglecting visual depth as a potential input dimension.
  • The "Midas touch" problem exists, where intentional gaze inputs cannot be distinguished from accidental gaze fixation.
  • Using gaze depth as an interaction input faces usability and learnability challenges, particularly at greater distances where visual depth estimation accuracy and stability decrease.

• Research Motivation and Significance:

  • The natural depth adjustment behavior of the human eye provides an intuitive yet underutilized input dimension that could enhance gesture interaction or multimodal systems.
  • Aiming to address the "Midas touch" problem while improving interaction intuitiveness and learning efficiency.

• Related Work:

  • Previous studies explored gaze depth applications for selecting virtual objects or depth layers, but these solutions generally lack effective user training processes and stable visual depth control mechanisms.

Solution

• Method and Core Innovations:

  1. FocusFlow Method: Proposes a hands-free interaction method based on gaze depth, leveraging human visual depth perception for VR interaction design.
  2. Hierarchical User Interface: Organizes interactive content into transparent "virtual windows" at different depth layers, dynamically adjusting visibility based on the user's gaze depth.
  3. Learning Process and Visual Cues: Designs two learning strategies (progressive learning and adaptive learning) with visual cues to help users master gaze depth control and gradually build muscle memory.

• Implementation Steps and Key Technologies:

  1. Gaze Depth Detection:
     • Utilizes eye tracking functionality in existing VR headsets to calculate gaze depth.
     • Applies a moving average algorithm to denoise depth fluctuations, significantly improving signal usability.
  2. Interface Design and Interaction Logic:
     • Introduces "virtual window" interaction components, allowing users to activate windows by adjusting gaze depth, enabling target selection, quick preview, and safe activation functions.
  3. Adaptive Learning Strategies:
     • Guides users in perceiving and adjusting gaze depth through dynamically changing transparency visual markers.
     • Provides feedback mechanisms and gradually reduces depth guidance, eventually removing cues to help users develop natural depth control.

Research Results

• Specific Outcomes:

  • Proposed a hierarchical UI design supporting gaze depth interaction, addressing the "Midas touch" problem.
  • Developed three visual cues and two learning strategies, demonstrating users' ability to quickly master gaze depth interaction methods.
  • Successfully implemented hands-free target selection, quick preview, and safe activation functions.

• Experiments and Evaluation Results:

  • In a user study with 24 participants, users quickly learned gaze depth interaction methods:
    • The adaptive learning strategy improved users' proficiency in transitioning to cue-free operations, reducing error rates and shortening activation time to 1.3 seconds.
  • Compared to traditional dwell-time methods, FocusFlow achieved higher interaction efficiency, with a reduced false activation rate of approximately 5%.
  • However, gaze depth interaction may lead to more noticeable eye fatigue, especially during prolonged and frequent use.

• Advantages Comparison:

  • FocusFlow demonstrated superior performance in efficiency (shorter average activation time) and false activation rate (lower than traditional methods), while addressing the "Midas touch" problem in a natural and intuitive manner.
  • Compared to traditional methods, FocusFlow supports a freer observation experience, avoiding interruptions caused by dwell-based operations.

• Limitations and Future Directions:

  1. Statistical Significance Analysis: The current sample size is small; larger-scale user studies are needed for long-term evaluation to verify broader applicability.
  2. Multimodal Expansion: Explore the integration of visual depth with other inputs (e.g., blinking, gestures).
  3. Continuous Input Dimension: Investigate the potential of gaze depth changes as a continuous input dimension (e.g., page scrolling, 3D object rotation).
  4. Dynamic Depth Adaptation: Dynamically adjust virtual window depth based on the distance of target objects to enhance user experience.
  5. Broader Application Scenarios: Introduce "perspective" applications (e.g., multi-layer observation in medical training) and other new interaction mechanisms.

Conclusion

FocusFlow introduces an innovative gaze depth-based interaction method, transforming the natural characteristics of the human visual system into intuitive input mechanisms in virtual reality. Through theoretical analysis, user studies, and experimental validation, this method demonstrates significant potential, particularly in addressing challenges in traditional gaze interactions. This work provides a theoretical foundation and direction for future research, further promoting this user-friendly interaction approach.