NVIDIA and Hugging Face have expanded the open-source LeRobot platform with Isaac GR00T 1.7 and the open Isaac Teleop framework, giving developers a standardised workflow to build, train and deploy humanoid robots more efficiently.
NVIDIA has expanded its collaboration with Hugging Face by bringing its Isaac GR00T 1.7 humanoid AI model and the open-source Isaac Teleop framework to LeRobot, strengthening the open-source robotics ecosystem with a more standardised path for robot development.
LeRobot, Hugging Face’s open-source robotics platform, now integrates Isaac GR00T 1.7, NVIDIA’s vision-language-action (VLA) foundation model for humanoid robots, alongside Isaac Teleop, an open-source framework that enables high-quality robot training through human demonstrations. Together, the additions provide developers with a streamlined workflow spanning data collection, model training, testing and deployment.
The collaboration is designed to accelerate physical AI development through open collaboration while lowering longstanding robotics barriers such as expensive data collection, simulation, computing requirements and validation. NVIDIA said the partnership connects its community of more than three million robotics developers with Hugging Face’s 16 million AI developers.
NVIDIA also plans to integrate Cosmos 3 into LeRobot, enabling synthetic robotics data generation, environment simulation and robot policy development when collecting real-world data is difficult or costly.
Isaac GR00T 1.7 is described by NVIDIA as its first open, commercially viable robot foundation model, designed to simplify post-training, deployment and adaptation across different robot designs.
The expanded platform also builds on existing open resources, including an open physical AI dataset with more than 350,000 real and simulated robot trajectories, 57 million grasp examples, and support for Isaac Sim, Isaac Lab, Isaac Lab-Arena, NVIDIA Jetson Thor and Reachy 2 humanoid robot deployments.
