Bi3: A Biplatform, Bicultural, Biperson Dataset for Social Robot Navigation
Published in IEEE International Conference on Robotics and Automation (ICRA) 2026, 2026
Recommended citation: Andrew Stratton, Phani Teja Singamaneni, Pranav Goyal, Rachid Alami, Christoforos Mavrogiannis, "Bi3: A Biplatform, Bicultural, Biperson Dataset for Social Robot Navigation." In the proceedings of IEEE International Conference on Robotics and Automation (ICRA), 2026. https://arxiv.org/abs/2605.06863
A dataset of close social robot navigation encounters between two humans and a robot, collected across two robot platforms and two sites (USA and France) with 74 participants. Bi3 provides 10.5 hours of multimodal human/robot motion tracks, RGB video, and user impressions, offering a benchmark for human motion prediction and robot control policies in densely occupied spaces. Code and data: github.com/fluentrobotics/bi3dataset.