Related Experiment Video
Updated: Jan 14, 2026
05:30
Soft Pneumatic Robot Modulates Graph Theory Metrics of Brain Network for Hand Rehabilitation After Stroke
Published on: October 10, 2025
420
An approach for unsupervised interaction clustering in human-robot co-work using spatiotemporal graph convolutional
Aaron Heuermann1,2, Zied Ghrairi2, Anton Zitnikov1
1Faculty 04: Production Engineering - Mechanical Engineering and Process Engineering, University of Bremen, Bremen, Germany.
Frontiers in Robotics and AI
|October 17, 2025
Summary
This study introduces a new method using spatiotemporal graph convolutional networks (STGCNs) to cluster human-robot interaction forms in industrial settings. The approach identifies 10 distinct interaction types, enabling more flexible and human-centered Industry 5.0 systems.
Area of Science:
- Robotics and Automation
- Human-Computer Interaction
- Industrial Engineering
Background:
- Future workplaces will see increased human-robot collaboration, necessitating flexible interaction models.
- Existing human-robot interaction taxonomies may not capture the full complexity of industrial co-working.
- Industry 5.0 emphasizes human-centered, adaptable manufacturing systems.
Purpose of the Study:
- To develop and evaluate a method for clustering diverse human-robot interaction forms in industrial co-work.
- To identify granular interaction patterns beyond current classifications.
- To support data-driven analysis for enhanced human-robot system design.
Main Methods:
- Utilized spatiotemporal graph convolutional networks (STGCNs) for interaction clustering.
- Collected data from 12 realistic industrial human-robot co-work scenarios.
- Employed a high-accuracy tracking system for precise data acquisition.
Main Results:
- Successfully clustered human-robot interactions into 10 distinct forms.
- Revealed more granular interaction patterns compared to established taxonomies.
- Demonstrated the effectiveness of STGCNs for modeling complex interactions.
Conclusions:
- The proposed STGCN approach effectively models and clusters human-robot interactions in industrial settings.
- Identified interaction forms provide deeper insights into collaborative behaviors.
- Findings contribute to developing more adaptive, human-centered Industry 5.0 systems.
