NVIDIA unveils GR00T N1 model for humanoid robotics

NVIDIA has revealed new technologies including the Isaac GR00T N1, a highly customisable foundation model for humanoid robots, aimed at accelerating robot development and addressing global labour shortages.

The release includes simulation frameworks and tools such as the NVIDIA Isaac GR00T Blueprint and Newton, an open-source physics engine developed alongside Google DeepMind and Disney Research, designed to bolster robotic advancements.

The GR00T N1 model is offered as a foundational tool that developers globally can pre-train for various robotic applications. "The age of generalist robotics is here," said Jensen Huang, Founder and CEO of NVIDIA. "With NVIDIA Isaac GR00T N1 and new data-generation and robot-learning frameworks, robotics developers everywhere will open the next frontier in the age of AI."

The GR00T N1 model operates with a dual-system architecture inspired by human cognition. "System 1" mirrors intuitive actions akin to human reflexes, while "System 2" allows for slower, thoughtful decision-making aided by a vision language model. System 1 learns from human demonstration data and synthetic data produced through the NVIDIA Omniverse platform.

Capabilities offered by GR00T N1 are applicable across a range of tasks like grasping and moving objects, with potential for multistep operations. These functions are suitable for industries like material handling and inspection. Developers can further adapt the model using real or synthetic data specific to their robotic applications.

In a demonstration, a humanoid robot from 1X completed domestic tasks using GR00T N1, showcasing NVIDIA's collaboration with other developers. "The future of humanoids is about adaptability and learning," said Bernt Børnich, CEO of 1X Technologies. "NVIDIA's GR00T N1 model provides a major breakthrough for robot reasoning and skills. With a minimal amount of post-training data, we were able to fully deploy on NEO Gamma — furthering our mission of creating robots that are not tools, but companions that can assist humans in meaningful, immeasurable ways."

Leading robotics companies such as Agility Robotics, Boston Dynamics, Mentee Robotics, and NEURA Robotics are amongst the early adopters of GR00T N1.

NVIDIA, Google DeepMind, and Disney Research have collaborated on Newton, a new physics engine aimed at enhancing robotic task complexity handling. The engine leverages the NVIDIA Warp framework, ensuring compatibility with Google DeepMind's MuJoCo and NVIDIA Isaac Lab systems.

MuJoCo-Warp is a collaborative venture between Google DeepMind and NVIDIA, targeting a significant increase in the efficiency of robotic machine learning workloads. Disney Research, a collaborator in this development, showcased its application through entertainment robots like the BDX droids.

Kyle Laughlin, Senior Vice President at Walt Disney Imagineering Research & Development, stated, "The BDX droids are just the beginning. We're committed to bringing more characters to life in ways the world hasn't seen before, and this collaboration with Disney Research, NVIDIA and Google DeepMind is a key part of that vision. This collaboration will allow us to create a new generation of robotic characters that are more expressive and engaging than ever before — and connect with our guests in ways that only Disney can."

To tackle the challenges in generating large datasets necessary for training humanoid robots, NVIDIA introduced the Isaac GR00T Blueprint for producing synthetic motion data. Running on Omniverse and NVIDIA Cosmos Transfer models, the tool significantly boosts data generation from limited human demonstrations.

In a recent demonstration, NVIDIA produced 780,000 synthetic data trajectories within 11 hours, substantially augmenting real-world data and enhancing GR00T N1's performance by 40%. Additionally, the GR00T N1 dataset is part of a larger open-source AI dataset now available to developers.

The GR00T N1 model, training data, and evaluation scenarios are accessible through platforms like Hugging Face and GitHub. Meanwhile, the soon-to-be-released Newton physics engine is expected to further aid robotic development efforts.