Modeling the Impact of Visual Stimuli on Redirection Noticeability with Gaze Behavior in Virtual Reality

Research Background and Issues

• Issues and Challenges:
  The authors explore the issue of "noticeability" of virtual avatar motion redirection in Virtual Reality (VR). While users' virtual actions can be redirected to enable innovative interactions, excessive redirection may lead to users noticing the deviation, thereby disrupting their sense of "embodiment" with the virtual avatar.
  Current research primarily focuses on the impact of virtual avatar properties (e.g., redirection magnitude and direction) on noticeability, but the influence of visual stimuli on noticeability through users' behavioral patterns remains underexplored. Moreover, the diversity of visual stimuli and individual differences in response pose challenges for quantifying and modeling their effects.

• Significance:
  Understanding and controlling the noticeability of redirection in VR is crucial for enhancing users' sense of immersion with virtual avatars, optimizing interaction performance, and advancing the practical applications of VR technology.

• Research Motivation:
  The authors propose capturing users' responses to visual stimuli through "gaze behaviors" to predict the noticeability of redirection. By complementing existing work, this study investigates the impact of visual stimuli in the environment on redirection noticeability and provides support for designing dynamic adjustment strategies.

Solution

• Methods and Key Contributions:
  The authors developed a computational model based on users' gaze behaviors to predict the noticeability of redirection under different visual stimulus conditions.

  • Visual responses were measured using gaze data (e.g., gaze point location, saccadic behavior, gaze duration, pupil activity index (IPA)).
  • User experiments were conducted in controlled environments to establish the relationship between gaze features and noticeability, leading to the development of regression and classification models.

• Innovations:

  1. Proposed using users' gaze behaviors (e.g., pupil activity, saccade rate, gaze duration) to indirectly quantify the impact of visual stimuli.
  2. Developed an accurate regression model capable of real-time adaptation to dynamically changing visual events.
  3. Applied the model to adaptive redirection techniques and validated its practicality in real-world scenarios.

• Implementation Steps:

  1. Confirm Research Methodology: Conduct preliminary studies to verify whether visual stimuli significantly affect redirection noticeability and confirm the correlation between gaze behaviors and noticeability.
  2. Data Collection: Design dual-task experiments to collect data, control the type and intensity of visual stimuli, and record gaze behaviors and noticeability outcomes.
  3. Model Development: Select the most effective combination of gaze features and construct a model based on Support Vector Regression (SVR).
  4. Model Validation and Evaluation: Test the model's predictive performance under new visual stimuli and user scenarios.
  5. Real-world Application Testing: Design two real-world scenarios (a VR action game and a boxing training simulation), apply dynamic adjustment techniques, and conduct user experience evaluations.

Research Outcomes

• Specific Results:

  1. The developed regression model can accurately predict noticeability under different visual stimulus conditions based on gaze behavior data, achieving a mean squared error (MSE) of 0.011.
  2. The model demonstrated good generalization performance with unseen visual stimuli and new users (e.g., MSE of 0.012 and 0.014 for color and scaling animation stimuli, respectively).

• Advantages:

  • The model supports dynamic redirection adjustments, significantly reducing users' physical burden while enhancing their sense of embodiment and control.
  • It is more intelligent and user-friendly compared to static redirection schemes.

• Experimental or Evaluation Results:

  1. User experiments showed that under visual stimulus interference, users' noticeability of redirection significantly decreased, and the paradigm and intensity of visual stimuli dynamically influenced noticeability.
  2. Preliminary validation of adaptive redirection techniques indicated that, compared to static redirection, users reported reduced physical burden and enhanced embodiment.

• Limitations and Future Directions:

  1. The experiments used relatively abstract visual stimuli; future work should incorporate more complex and dynamic visual events.
  2. The range of rotation angles and arm postures was limited; future studies could explore the impact of redirection on different body parts.
  3. Personalized modeling is recommended to further improve model performance, such as integrating physiological signals (e.g., heart rate, EEG) to build a richer behavioral feature set.
  4. Investigating long-term adaptation to redirection and users' sustained perception of redirection is an important research direction.

Conclusion

This study advances the understanding of how visual stimuli in VR environments affect users' noticeability of redirection, develops a modeling approach with strong generalization capabilities, and demonstrates its potential applications in real-world scenarios. This work provides significant theoretical and practical support for designing efficient and immersive VR interactions while offering directions for future research.