New paper in Sensors journal: "Towards a Safe Human-Robot Collaboration Using Information on Human Worker Activity"

This article explores a computational model for intuitive and adaptive human–robot interaction (HRI) in industrial settings, enabling safe and flexible collaboration. The authors propose a human-skeleton-based model using LSTM networks to recognize spatiotemporal worker activities, achieving 91.365% training accuracy on the InHARD dataset. By combining upper body positions with actions, the system enhances context-aware collaboration, allowing the robot to adapt its behavior to dynamic environments.

Authors: Orsag, Luka ; Stipančić, Tomislav ; Koren, Leon

DOI: 10.3390/s23031283