Noitom releases hiphi dataset for humanoid robotics and digital human research

Noitom Robotics unveiled HiPHI, a high-precision optical motion capture dataset designed for researchers and engineers in humanoid robotics, digital humans, and computer graphics. The dataset spans 617.5 hours, including 371.8 hours of full-body human motion and 245.7 hours of human-object interaction data, with mirrored versions. It was collected from 132 motion capture actors using 90 Hz sampling and sub-millimeter optical accuracy, structured according to the FrameNet framework for action semantics. HiPHI is now publicly available on Hugging Face, accompanied by documentation and research papers. Models trained on HiPHI have been deployed on the Yuji G1 humanoid robot, enabling tasks such as running, sitting, crawling, carrying boxes, and pulling luggage. The dataset aims to advance the development of human-like movement in robotics and digital human technologies.