Enhancing Blind Visitor's Autonomy in a Science Museum Using an Autonomous Navigation Robot
Seita Kayukawa, Daisuke Sato, Masayuki Murata, Tatsuya Ishihara, Hironobu Takagi, Shigeo Morishima, Chieko Asakawa · 2023 · Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3581220
Summary
Museums have historically struggled to offer blind visitors the same self-directed, exploratory experience that sighted visitors take for granted. Prior solutions — tactile tours, audio guides, or pre-fixed navigation routes — either require a human escort or constrain visitors to scripted paths that remove choice and pacing. This paper reports on a system designed at Miraikan (The National Museum of Emerging Science and Innovation, Tokyo) that pairs a suitcase-shaped autonomous navigation robot with an iOS smartphone app and on-demand support from museum Science Communicators (SCs). The robot, built on the open-source CaBot platform, uses 3D LiDAR and an RGB-Depth camera with YOLOv4 object detection to avoid obstacles and other visitors; users grip a haptic handle and follow the robot while listening to short audio descriptions of upcoming exhibits. The smartphone app, fully compatible with VoiceOver, lets the visitor select destinations, browse long-form exhibit descriptions at a self-chosen reading speed, and summon an SC through a Google Cloud Firestore back end when deeper interaction is wanted. The authors ran a real-world study during the museum's regular opening hours with eight totally blind participants (P1–P8, ages 25–53) exploring a 2,100 m² floor for 90 minutes each, alongside a 108-person questionnaire measuring sighted visitors' social acceptance and two follow-up focus groups on remaining needs.
Key findings
All eight blind participants completed an independent, self-paced museum visit with no safety incidents despite ~272 other visitors circulating each day. Every participant rated the experience of walking with the robot at 7/7 (enjoyment) and being able to choose their own pace at 6–7/7; six of eight went beyond the predefined 10-exhibit tour to revisit exhibits of personal interest. On average participants walked with the robot for 9 minutes, used the smartphone app for 17 minutes, and spent 26 minutes (30%) of their time exploring without any human assistance — a balance that participants described as considerably more autonomous than prior museum visits. Among the 108 sighted bystanders, 99.1% agreed assistive robots should be introduced in museums, only 2.8% found the blind-person/robot pair disruptive, 86.1% perceived no safety risk, and 78.7% accepted the robot's outward-facing camera even without data-retention guarantees — a notably permissive result for camera-based assistive technology. Focus groups surfaced a consistent future requirement: participants want finer-grained navigation inside each sub-exhibit (not just to the entrance) and prefer calling an SC only when they have specific questions, rather than being escorted continuously.
Relevance
For practitioners designing public-space accessibility, this study is an existence proof that a blind visitor can move through a busy, uncontrolled environment with the dignity of a self-directed tourist rather than a dependent. It shifts the conversation from whether autonomous assistive robots are socially acceptable (they were) to how the division of labour between machine navigation and human interpretation should be designed. The system's split — robot handles mobility and wayfinding, human SCs handle interpretation and Q&A — is a pragmatic template for museums, airports, and transit hubs. Limitations are worth naming: the study ran during post-COVID low-traffic periods (272 vs. the pre-pandemic 4,000–5,000 daily visitors), eight participants is a small sample, and the robot guided users only between sub-exhibits, not within them. Still, the finding that even camera privacy concerns were broadly tolerated suggests deployment barriers may be lower than assumed.
Tags: blind navigation · autonomous navigation robot · museum accessibility · visual impairment · indoor navigation · assistive robotics · social acceptance · wayfinding · LiDAR · human-robot interaction