New Study on Human-Robot Interaction Reveals Insights on Proxemics in Crowded Spaces

A recent large-scale empirical study published in Science Advances by teams from ETH Zürich, SPF, and EPFL examined human-robot interaction (HRI) in crowded environments. Utilizing four real-world datasets across Europe, North America, and Asia, the research established a benchmark for social navigation distances based on density perception and speed dependency.

This study is significant as it addresses the ongoing challenge of how robots can comfortably share space with humans, a concept rooted in proxemics. Previous research indicated that pedestrians maintain specific distances from each other, but robots behave differently. The findings suggest that pedestrians tend to give robots more space, which varies with the robot's speed and the density of the crowd.

Looking ahead, the CrowdBot dataset, which includes various robot platforms operating at different speeds, will serve as a foundation for future studies. The research team has developed a unified analysis pipeline to quantify behaviors based on movement dynamics and proxemics, paving the way for improved human-robot interactions in public spaces. No further timeline was disclosed at the time of publication.