本文由 AI 分析生成
建立時間： 2026-04-02 來源： https://arxiv.org/abs/2602.23475
Summary
Using video analysis of cleaning robots deployed in a commercial airport, this paper documents how technically proficient robots disrupt the social order of transit spaces because they lack three core interactional competencies: mutual adjustment to others, understanding of social groups, and awareness of location purposes. The authors propose “socially-aware movement” as a design space and develop “strong concepts” that treat movement as a collaborative, interactional accomplishment.
透過分析商業機場清潔機器人的影片，本文記錄機器人如何因缺乏三項互動能力而擾亂公共空間社會秩序：對他人的相互調適、理解社會群體、以及感知地點目的。作者提出「社會感知移動」作為設計空間。
Prerequisites
- Conversation analysis (CA) and ethnomethodology — the methodology is video analysis in the CA tradition; “troubles” as an analytic resource is a CA concept.
- Robot navigation and autonomy — the robots studied are technically functional (they navigate without collisions); the failures are social, not technical, which requires understanding what current navigation systems do and don’t model.
- “Strong concepts” in HCI design — the authors use this theoretical framework to generalize findings into reusable design principles; familiarity with HCI theory-building helps.
Core Idea
Current robot navigation optimizes for obstacle avoidance and path efficiency but treats other humans as obstacles rather than social agents. In practice, human movement in public spaces is deeply cooperative: people negotiate passing, adjust for groups, and modulate pace based on location context (rushing through a corridor vs. browsing a terminal). Cleaning robots violate these unwritten norms constantly — cutting through groups, failing to yield when humans signal a desire to pass, misreading high-dwell areas as passable. The authors argue these aren’t edge cases but fundamental gaps, and propose that “movement as interaction” should be a first-class design concern.
Results
- Video corpus: publicly deployed cleaning robots in a major commercial airport.
- Key failure categories documented: (1) not yielding to mutual adjustment signals, (2) disrupting social groups (families, pairs), (3) inappropriate behavior in functionally distinct locations.
- Qualitative findings — no quantitative metrics reported.
Limitations
- Author-stated: single deployment context (airport); findings may not generalize to other transit spaces (hospitals, shopping centers).
- Unstated: cleaning robots are a specific morphology (low, disk-shaped); social navigation challenges may differ for humanoid or taller robots that make eye contact easier.
Reproducibility
- Code: N/A (video analysis study).
- Datasets: video corpus from airport deployment; privacy constraints likely prevent public release.
- Compute: N/A.
Insights
This is a rare study that observes deployed robots (not lab setups) and analyzes real social failures. The finding that robots “disrupt the social order” even when technically correct is important: the bar for public-space robots isn’t just collision-free navigation but socially legible navigation. The “strong concepts” framing — treating movement as interactional — is a useful design vocabulary that maps onto recent work in socially-aware navigation (SAN) and proxemics-aware planning.
Connections
Raw Excerpt
These robots, while technically proficient, can disrupt the social order of a space due to their inability to understand core aspects of human movement: mutual adjustment to others, the significance of understanding social groups, and the purpose of different locations.