Privacy Risks of Robot Vision: A User Study on Image Modalities and Resolution

2025-05-12

AI Summary

This study investigates how image modalities (RGB, depth, semantic segmentation) and spatial resolution (16×16, 32×32) influence users’ perceived privacy risk in service robot vision systems. Method: Through a large-scale controlled experiment with Likert-scale privacy assessments, we quantitatively evaluated user perceptions across modalities and resolutions. Contribution/Results: We provide the first empirical evidence that depth and semantic segmentation images are perceived as significantly lower-risk than RGB imagery. A 32×32 RGB image was rated as “nearly sufficient for privacy protection,” while 16×16 RGB was deemed “fully privacy-safe” by most participants. These findings challenge the RGB-centric privacy assumptions prevalent in robotics and establish low-resolution imaging and non-RGB modalities as effective, lightweight, and deployable privacy-enhancing strategies. The results offer empirically grounded, quantitative guidance for designing robot vision systems that balance functional performance with user trust and privacy assurance.

Technology Category

• Computer Vision: Bias, Fairness & Privacy
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Natural Language Processing: Safety and Robustness

Application Category

• Security and Privacy: Data transparency and provenance
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

User privacy is a crucial concern in robotic applications, especially when mobile service robots are deployed in personal or sensitive environments. However, many robotic downstream tasks require the use of cameras, which may raise privacy risks. To better understand user perceptions of privacy in relation to visual data, we conducted a user study investigating how different image modalities and image resolutions affect users' privacy concerns. The results show that depth images are broadly viewed as privacy-safe, and a similarly high proportion of respondents feel the same about semantic segmentation images. Additionally, the majority of participants consider 32*32 resolution RGB images to be almost sufficiently privacy-preserving, while most believe that 16*16 resolution can fully guarantee privacy protection.

Problem

• Investigates privacy risks of robot vision in sensitive environments
• Examines user perceptions of privacy across image modalities
• Evaluates impact of image resolution on privacy concerns

Innovation

• Depth images ensure privacy safety
• Semantic segmentation images reduce privacy concerns
• Low resolution RGB images preserve privacy