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

With the rapid advancements in robotics, mobile service robots have become increasingly essential in assisting people with a wide variety of tasks, including domestic chores, healthcare, and package delivery [1, 2, 3]. To efficiently accomplish these tasks, most mobile robots are equipped with high-resolution cameras that capture detailed visual data of their operational environments. Although the usage of these visual sensors enhances robot performance, it simultaneously raises substantial privacy concerns [6], particularly when robots operate within users' personal or private spaces. Motivated by the critical need to balance robot performance and user privacy, it is important to understand user perceptions of privacy concerns related to robotic visual data collection. Previous studies have investigated privacy concerns related to general camera surveillance [4, 5], yet relatively few studies specifically focus on visual data modalities and image resolution for privacy in the context of mobile robotics. Therefore, we conducted a user study aimed at uncovering user preferences and attitudes regarding privacy risks associated with robotic visual perception. Specifically, our objectives include evaluating user opinions on privacy of different visual data modalities and determining user-preferred strategies and thresholds (e.g., reduced image resolution) for effective privacy preservation.