NVIDIA and Hugging Face Unite to Advance Open Source Robotics with LeRobot Integrations

Open Source Robotics: Unlocking Physical AI  - Open source AI has proven how quickly innovation can accelerate when models, data, and tools are shared. Robotics now faces the same opportunity — but progress in physical AI is often slowed by fragmented and costly resources, from large datasets and robot foundation models to simulation, compute, and validation frameworks.

To address this, NVIDIA and Hugging Face are partnering to bring powerful robotics tools into the open. The collaboration integrates the NVIDIA Isaac GR00T 1.7 reasoning vision-language-action (VLA) model for humanoid robots and the NVIDIA Isaac Teleop framework directly into LeRobot, Hugging Face’s open source robotics library. Soon, NVIDIA Cosmos 3, a frontier world model for physical AI, will also join the ecosystem. Together, these integrations create a more accessible, standardized pipeline for end-to-end robot development, fueling collaboration across the global robotics community.

Thomas Wolf, cofounder and chief science officer at Hugging Face, explains: “Open source is how advanced research becomes something people can study, adapt, and build on. With NVIDIA Isaac GR00T 1.7 and Isaac Teleop in LeRobot today, developers can train and evaluate robots in the open. And with NVIDIA Cosmos 3 planned next, the community will be able to bring frontier world models into that same collaborative loop.”

A Unified Pipeline for Robot Foundation Models

LeRobot is an open source library for training, running, and sharing robot datasets, models, policies, and workflows. By connecting NVIDIA’s 3 million robotics developers with Hugging Face’s 16 million AI builders, this partnership expands access to frontier physical AI tools through open, interoperable workflows.

Key integrations include:

• NVIDIA Isaac Teleop
  An open source framework for robot data collection. It enables developers to capture high-quality human demonstrations from external devices in standardized formats, expand datasets, and share them directly within LeRobot.
• NVIDIA Isaac GR00T 1.7
  The first open, commercially viable robot foundation model. It simplifies post-training and deployment through LeRobot workflows, allowing developers to adapt GR00T to new robot embodiments and tasks with benchmarked performance.
• NVIDIA Cosmos 3 (coming soon)
  A frontier world foundation model for physical AI. It will help generate and augment robotics data, simulate scenarios, and support policy development when real-world data is limited or costly.

Building on NVIDIA’s Broader Ecosystem

These integrations extend a suite of NVIDIA resources already connected to LeRobot:

• Largest open source physical AI dataset
  Over 350,000 real and simulated trajectories and 57 million grasps, downloaded more than 15 million times, to accelerate robotics workflows.
• NVIDIA Isaac Sim and Isaac Lab
  Simulation frameworks for environment setup, data generation, policy testing, and behavior validation before deployment on physical robots.
• Isaac Lab-Arena in LeRobot EnvHub
  Enables rapid prototyping of complex simulation environments, registration in EnvHub, and seamless use within LeRobot to train and evaluate generalist robot policies such as GR00T, Pi, and SmolVLA.
• NVIDIA Jetson Thor integration with LeRobot’s Reachy 2
  Supports deployment of VLA models on open source humanoid robots.

Learn More

Explore how to use Isaac Teleop, Isaac GR00T 1.7, and Isaac Lab-Arena with LeRobot for end-to-end humanoid development, and dive into detailed LeRobot integration workflows to accelerate your robotics journey.