What You Experience is What We Collect: User Experience Based Fine-Grained Permissions for Everyday Augmented Reality

Title of the Paper

What You Experience is What We Collect: User Experience Based Fine-Grained Permissions for Everyday Augmented Reality

Paper Information

• Domain: Privacy and permission control in augmented reality devices
• Keywords: AR permissions, privacy protection, sensor data management, user experience, fine-grained permissions, augmented reality perception

Research Background and Problem

• Identified Problems or Challenges: Traditional binary permission control methods are inadequate for the always-on nature of augmented reality devices. This can lead to a lack of user understanding regarding the scope of sensor data access, resulting in privacy risks.
• Significance: Augmented reality devices involve a broader range of sensor data with higher sensitivity, necessitating a more transparent permission control design to help users better understand the trade-offs between privacy and functionality.
• Research Motivation and Related Work:
  • Current permission systems are primarily designed for smartphones and are not well-suited for always-on AR devices.
  • Previous studies indicate that users often lack understanding of permission requests and their reasons, leading to lower trust levels.
  • The "always-on" state of sensors in AR devices raises privacy concerns, requiring new approaches to address these issues.

Proposed Solution

• Proposed Solution: Develop a user experience-based fine-grained permission system that allows users to adjust data access granularity via sliders while showing how application functionality is affected by changes in data provision.
• Innovations: Combining textual descriptions with application images to dynamically present the specific impact of data access restrictions on application functionality, enhancing the flexibility and transparency of user permissions.
• Implementation Steps and Techniques:
  • Permission Control Methods: Includes slider-based control with and without images, compared against current mainstream binary control (Binary), Android 11, and iOS permission systems.
  • Experimental Design:
    1. Test permission control methods in five application scenarios (e.g., interior design, health tracking, etc.);
    2. Allow users to set data access levels via sliders and observe functionality changes;
    3. Evaluate results using quantitative metrics such as NASA-TLX workload, user comprehension, and trust levels.
  • Technical Implementation: Developed using Unity and deployed on AR devices, with users operating the permission control interface via a handheld controller.

Research Outcomes

• Specific Findings:
  • Proposed and evaluated five permission control methods.
  • Demonstrated that the slider-based control with images performed best in improving user understanding of permission requests and privacy impacts, significantly enhancing overall comprehension of privacy trade-offs.
  • Experiments showed that users preferred the combination of images and text in the slider control method, with significantly higher usability and trust compared to traditional approaches.
  • Users made more rational permission choices, customizing data access levels based on privacy and functionality needs.
• Advantages:
  • The slider control method offers a more intuitive way to adjust permissions compared to existing systems.
  • Significantly improved user awareness of application data usage and privacy risks.
  • Enhanced transparency increased user trust in devices and applications while promoting acceptance of fine-grained permissions.
• Experimental or Evaluation Results:
  • The slider with images method had the highest preference rate among users (80%, ranked first).
  • The slider with images method excelled in information accuracy, privacy decision opportunities, and trust levels.
  • Traditional binary permission control systems ranked lowest in user understanding, trust, and usability.
• Limitations and Future Directions:
  • Limitations:
    • Small sample size (N=20); further studies with larger samples are needed to validate generalizability.
    • Slider control images may be subject to developer bias or distortion, requiring additional verification mechanisms.
  • Future Directions:
    • Investigate the adaptability of the slider permission system in specific application scenarios.
    • Explore dynamic integration of user privacy preferences with real-world usage contexts, such as automatic location- or session-based permission decisions.
    • Extend the slider permission method to real-time data stream control, incorporating the "principle of least privilege."

This summary highlights the key findings and contributions of the paper, focusing on technical advancements and their implications for privacy and usability in augmented reality environments.