Signal or 'Noise': Human Reactions to Robot Errors in the Wild

TL;DR

Robots frequently err in real-world settings, and human social responses to those errors—especially for non-social robots in group and repeated interactions—are not well understood. The authors built an autonomous coffee-robot deployment to study these reactions in the wild, collecting 39 complete interactions across 49 participants over three weeks. They find that social signals are information-rich and frequently volunteered but also noisy, with group interactions producing higher signal density and more noise. The results offer guidelines for context-aware interpretation and integration of social cues into real-world HRI systems, highlighting opportunities and challenges for error management in the wild.

Abstract

In the real world, robots frequently make errors, yet little is known about people's social responses to errors outside of lab settings. Prior work has shown that social signals are reliable and useful for error management in constrained interactions, but it is unclear if this holds in the real world - especially with a non-social robot in repeated and group interactions with successive or propagated errors. To explore this, we built a coffee robot and conducted a public field deployment ($N = 49$). We found that participants consistently expressed varied social signals in response to errors and other stimuli, particularly during group interactions. Our findings suggest that social signals in the wild are rich (with participants volunteering information about the interaction), but "noisy." We discuss lessons, benefits, and challenges for using social signals in real-world HRI.

Figures (2)

• Figure 1: Coffee robot system physical setup (left) and interaction workflow (right).
• Figure 2: Coffee robot system diagram. Dashed lines and boxes show how an error management system using social signals could be integrated into it.