Nvidia, Hugging Face expand LeRobot with open robotics AI tools

Nvidia and Hugging Face have announced an expanded partnership to accelerate robot development through open-source tools and models. The integration brings two key Nvidia technologies into LeRobot, Hugging Face's open-source robotics library for developing, training, and sharing robot datasets, models, and workflows.

The centerpiece additions are Nvidia Isaac GR00T 1.7, an open vision-language-action (VLA) foundation model designed specifically for humanoid robots, and the Nvidia Isaac Teleop framework, which enables developers to collect human demonstrations from external devices using standardized formats. A third component—Nvidia Cosmos 3, a world foundation model for physical AI—is planned for future integration and will allow developers to generate synthetic robotics data, simulate environments, and develop robot policies when real-world data is unavailable or too costly to collect. Together, these tools provide a standardized workflow for the entire robot development cycle: collecting data, training models, evaluating performance, and deploying AI-powered systems.

Isaac GR00T 1.7 is designed to simplify the post-training and deployment of robot foundation models across different robot types and applications, removing barriers that have historically made robot customization labor-intensive. The expanded partnership builds on existing Nvidia resources already available through LeRobot, including open-source physical AI datasets containing more than 350,000 real and simulated robot trajectories and 57 million grasp samples, as well as simulation environments based on Nvidia Isaac Sim and Isaac Lab. Deployment support extends to Nvidia Jetson Thor running on Hugging Face's Reachy 2 humanoid robot, allowing developers to run vision-language-action models on open-source hardware platforms.

Thomas Wolf, cofounder and chief science officer at Hugging Face, framed the collaboration as a means of democratizing robotics research. "With Nvidia Isaac GR00T 1.7 and Isaac Teleop in LeRobot today, robotics developers can use shared models, data and workflows to train and evaluate robots in the open," he said, adding that "with Nvidia Cosmos 3 planned next, the community will have a path to bring frontier world models into that same collaborative loop." Nvidia estimates the partnership reaches more than three million robotics developers on its side, combining with Hugging Face's 16 million AI developers to form a significantly larger ecosystem for adoption of physical AI technologies.

Nvidia and Hugging Face are consolidating their respective developer ecosystems to accelerate adoption of physical AI—the application of AI models to real-world robotics tasks. LeRobot, Hugging Face's open-source robotics library, already provided a framework for developing and sharing robot datasets and models. By integrating Nvidia's foundation models and development frameworks, the partnership removes friction points in the robot development pipeline: data collection (via Isaac Teleop), model training and evaluation (via Isaac GR00T 1.7), and deployment (via Jetson Thor support on Reachy 2). The planned integration of Nvidia Cosmos 3, a world foundation model, will further extend this by allowing synthetic data generation and environment simulation—addressing one of robotics' most persistent challenges, the cost and difficulty of collecting diverse real-world training data.

The collaboration explicitly frames open source as the vehicle for translating research into accessible tools. Thomas Wolf emphasized that openness allows the field to "study, adapt and build on" advanced work. By combining Nvidia's three million robotics developers with Hugging Face's 16 million AI developers, the partnership broadens the addressable audience for physical AI and creates a shared ecosystem where models, datasets, and workflows can be reused across different robot types and applications.

No comments yet. Be the first to share your thoughts!