Thing Ethnography in a Factory: Exploring Emergent and Dynamic Relations of Cobots and Workers

Paper Title

Thing Ethnography in a Factory: Exploring Emergent and Dynamic Relations of Cobots and Workers

Publication Info

• Topic area: Human-robot collaboration in small-scale manufacturing environments.
• Keywords: Collaborative robots, cobots, thing ethnography, workplace automation, human-robot interaction, invisible labor, intra-action, agential realism, small-scale manufacturing, CNC machines.

Background and Problem

• Problem / challenge: Collaborative robots (cobots) are often assumed to operate as autonomous collaborators in industrial settings, but their real-world roles and dependencies on human workers remain underexplored. Existing research tends to overlook the dynamic and relational constitution of roles between cobots and human workers.
• Significance: Understanding the actual roles of cobots and their integration into production environments is crucial for improving workplace automation, addressing power dynamics, and supporting workers in increasingly automated settings.
• Motivation and related work: Prior studies have shown that cobots often perform repetitive tasks and require significant human intervention for maintenance and troubleshooting. Critiques from Science and Technology Studies (STS) and Human-Computer Interaction (HCI) highlight the invisible labor sustaining automation but fail to fully capture the relational dynamics between humans and robots. This paper builds on these insights by applying a relational lens to cobot integration.

Solution

• Proposed approach: A robot-centered thing ethnography conducted over 18 days in a small-scale motorcycle parts factory, using Barad’s concept of intra-action to analyze the co-constitution of roles between cobots, human workers, and other entities.
• Novelty:
  1. Introduces the concept of intra-action to analyze human-robot relations, moving beyond traditional interactional perspectives.
  2. Identifies three emergent relational configurations: cobots as frontline operators, care receivers, and scapegoating managers.
  3. Demonstrates the utility of thing ethnography as a methodology for uncovering invisible labor and relational dynamics in workplace automation.
• Procedure and key techniques:
  • Installed cameras to focus on cobots and their interactions with CNC machines and workers.
  • Collected 20,428 video clips over 18 days, analyzed using reflexive thematic analysis.
  • Applied Barad’s agential realism framework to identify and interpret relational configurations through agential cuts.

Results

• Concrete findings:
  • Cobots performed as frontline operators, relying on human workers for scheduling, quality control, and troubleshooting.
  • Human workers acted as caregivers, performing essential maintenance and environmental adjustments to sustain cobot operations.
  • Cobots, in conjunction with the factory owner and monitoring systems, acted as scapegoating managers, shifting responsibility for errors onto human workers.
• Advantage over baselines: Challenges the simplistic view of cobots as autonomous collaborators by revealing their dependency on human labor and the complex power dynamics in automated workplaces.
• Experiments / evaluation:
  • Conducted in a small factory with fewer than five workers, using CNC machines paired with cobots.
  • Data collected through event-triggered video recordings and analyzed to trace human-robot interactions and relational dynamics.
• Limitations and future work:
  • Lack of direct interviews with workers limits the ability to capture their subjective perspectives.
  • Future work could explore additional workplace contexts, incorporate worker voices, and develop design interventions to address power asymmetries.

Summary

This study uses an 18-day thing ethnography in a small factory to analyze the relational dynamics between cobots, human workers, and other entities. By applying Barad’s intra-action framework, the authors identify three emergent roles for cobots: frontline operators, care receivers, and scapegoating managers. These findings challenge traditional views of cobots as autonomous collaborators and highlight the invisible labor sustaining automation. The study contributes a novel methodological and theoretical lens for understanding workplace automation and offers actionable insights for HCI and HRI research on supporting workers in automated environments.