CMR Surgical (CMR), the global surgical robotics company, announced its participation in NVIDIA’s Physical AI healthcare robotics initiative, unveiled at NVIDIA GTC. As part of the initiative, CMR contributed the majority of surgical data used to create Open-H, the world’s largest open dataset for healthcare robotics, designed to train the next generation of intelligent surgical systems.

The dataset combines real-world surgical video, robotic telemetry and multimodal data from leading healthcare and robotics organisations. CMR contributed close to 500 hours of anonymised surgical data from its Versius Surgical Robotic System, representing the largest share of surgical data in the initiative.

Open‑H underpins Isaac GR00T‑H, the first open vision‑language‑action model for healthcare robotics, designed to enable robotic systems to better interpret complex surgical environments and tasks. These technologies aim to accelerate the development of intelligent robotic systems while maintaining the rigorous safety and clinical oversight required in healthcare.

Robotic surgery has already enabled millions of minimally invasive procedures worldwide, helping surgeons perform complex operations with greater precision and control. However, advancing surgical robotics further requires new approaches to how robotic systems learn from clinical experience.

With NVIDIA Physical AI infrastructure, the ecosystem has a platform that allows robotic systems to be trained and evaluated in simulated environments before deployment, helping accelerate development while maintaining high safety standards.

CMR is contributing real-world surgical data to the effort and is also using NVIDIA Cosmos-H to generate physically accurate synthetic surgical data and evaluate new robotic policies for the future development of the Versius platform.