Open-H-Embodiment: A Large-Scale Dataset for Enabling Foundation Models in Medical Robotics
Abstract
Autonomous medical robots hold promise to improve patient outcomes, reduce provider workload, democratize access to care, and enable superhuman precision. However, autonomous medical robotics has been limited by a fundamental data problem: existing medical robotic datasets are small, single-embodiment, and rarely shared openly, restricting the development of foundation models that the field needs to advance. We introduce Open-H-Embodiment, the largest open dataset of medical robotic video with synchronized kinematics to date, spanning more than 49 institutions and multiple robotic platforms including the CMR Versius, Intuitive Surgical's da Vinci, da Vinci Research Kit (dVRK), Rob Surgical BiTrack, Virtual Incision's MIRA, Moon Surgical Maestro, and a variety of custom systems, spanning surgical manipulation, robotic ultrasound, and endoscopy procedures. We demonstrate the research enabled by this dataset through two foundation models. GR00T-H is the first open foundation vision-language-action model for medical robotics, which is the only evaluated model to achieve full end-to-end task completion on a structured suturing benchmark (25% of trials vs. 0% for all others) and achieves 64% average success across a 29-step ex vivo suturing sequence. We also train Cosmos-H-Surgical-Simulator, the first action-conditioned world model to enable multi-embodiment surgical simulation from a single checkpoint, spanning nine robotic platforms and supporting in silico policy evaluation and synthetic data generation for the medical domain. These results suggest that open, large-scale medical robot data collection can serve as critical infrastructure for the research community, enabling advances in robot learning, world modeling, and beyond.
Keywords
Cite
@article{arxiv.2604.21017,
title = {Open-H-Embodiment: A Large-Scale Dataset for Enabling Foundation Models in Medical Robotics},
author = {Open-H-Embodiment Consortium and : and Nigel Nelson and Juo-Tung Chen and Jesse Haworth and Xinhao Chen and Lukas Zbinden and Dianye Huang and Alaa Eldin Abdelaal and Alberto Arezzo and Ayberk Acar and Farshid Alambeigi and Carlo Alberto Ammirati and Yunke Ao and Pablo David Aranda Rodriguez and Soofiyan Atar and Mattia Ballo and Noah Barnes and Federica Barontini and Filip Binkiewicz and Peter Black and Sebastian Bodenstedt and Leonardo Borgioli and Nikola Budjak and Benjamin Calmé and Fabio Carrillo and Nicola Cavalcanti and Changwei Chen and Haoxin Chen and Sihang Chen and Qihan Chen and Zhongyu Chen and Ziyang Chen and Shing Shin Cheng and Meiqing Cheng and Min Cheng and Zih-Yun Sarah Chiu and Xiangyu Chu and Camilo Correa-Gallego and Giulio Dagnino and Anton Deguet and Jacob Delgado and Jonathan C. DeLong and Kaizhong Deng and Alexander Dimitrakakis and Qingpeng Ding and Hao Ding and Giovanni Distefano and Daniel Donoho and Anqing Duan and Marco Esposito and Shane Farritor and Jad Fayad and Zahi Fayad and Mario Ferradosa and Filippo Filicori and Chelsea Finn and Philipp Fürnstahl and Jiawei Ge and Stamatia Giannarou and Xavier Giralt Ludevid and Frederic Giraud and Aditya Amit Godbole and Ken Goldberg and Antony Goldenberg and Diego Granero Marana and Xiaoqing Guo and Tamás Haidegger and Evan Hailey and Pascal Hansen and Ziyi Hao and Kush Hari and Kengo Hayashi and Jonathon Hawkins and Shelby Haworth and Ortrun Hellig and S. Duke Herrell and Zhouyang Hong and Andrew Howe and Junlei Hu and Zhaoyang Jacopo Hu and Ria Jain and Mohammad Rafiee Javazm and Howard Ji and Rui Ji and Jianmin Ji and Zhongliang Jiang and Dominic Jones and Jeffrey Jopling and Britton Jordan and Ran Ju and Michael Kam and Luoyao Kang and Fausto Kang and Siddhartha Kapuria and Peter Kazanzides and Sonika Kiehler and Ethan Kilmer and Ji Woong Kim and Przemysław Korzeniowski and Chandra Kuchi and Nithesh Kumar and Alan Kuntz and Federico Lavagno and Yu Chung Lee and Hao-Chih Lee and Hang Li and Zhen Li and Xiao Liang and Xinxin Lin and Jinsong Lin and Chang Liu and Fei Liu and Pei Liu and Yun-hui Liu and Wanli Liuchen and Eszter Lukács and Sareena Mann and Miles Mannas and Brett Marinelli and Sabina Martyniak and Francesco Marzola and Lorenzo Mazza and Xueyan Mei and Maria Clara Morais and Luigi Muratore and Chetan Reddy Narayanaswamy and Michał Naskręt and David Navarro-Alarcon and Cyrus Neary and Chi Kit Ng and Christopher Nguan and David Noonan and Ki Hwan Oh and Tom Christian Olesch and Allison M. Okamura and Justin Opfermann and Matteo Pescio and Doan Xuan Viet Pham and Tito Porras and Hongliang Ren and Ariel Rodriguez Jimenez and Ferdinando Rodriguez y Baena and Septimiu E. Salcudean and Asmitha Sathya and Preethi Satish and Lalithkumar Seenivasan and Jiaqi Shao and Yiqing Shen and Yu Sheng and Lucy XiaoYang Shi and Zoe Soulé and Stefanie Speidel and Mingwu Su and Jianhao Su and Idris Sunmola and Kristóf Takács and Yunxi Tang and Patrick Thornycroft and Yu Tian and Jordan Thompson and Mehmet K. Turkcan and Mathias Unberath and Pietro Valdastri and Carlos Vives and Quan Vuong and Martin Wagner and Farong Wang and Wei Wang and Lidian Wang and Chung-Pang Wang and Guankun Wang and Junyi Wang and Erqi Wang and Ziyi Wang and Tanner Watts and Wolfgang Wein and Yimeng Wu and Zijian Wu and Hongjun Wu and Luohong Wu and Jie Ying Wu and Junlin Wu and Victoria Wu and Kaixuan Wu and Mateusz Wójcikowski and Yunye Xiao and Nan Xiao and Wenxuan Xie and Hao Yang and Tianqi Yang and Yinuo Yang and Menglong Ye and Ryan S. Yeung and Nural Yilmaz and Chim Ho Yin and Michael Yip and Rayan Younis and Chenhao Yu and Sayem Nazmuz Zaman and Milos Zefran and Han Zhang and Yuelin Zhang and Yidong Zhang and Yanyong Zhang and Xuyang Zhang and Yameng Zhang and Joyce Zhang and Ning Zhong and Peng Zhou and Haoying Zhou and Xiuli Zuo and Nassir Navab and Mahdi Azizian and Sean D. Huver and Axel Krieger},
journal= {arXiv preprint arXiv:2604.21017},
year = {2026}
}
Comments
Project website: https://open-h.github.io/open-h-embodiment/