Nvidia unveils Isaac GR00T blueprint for accelerating humanoid robots

Nvidia Announce Isaac GR00T A scheme to accelerate the development of humanoid robots.

in Consumer Electronics Show 2025 In Nvidia CEO Jensen Huang’s keynote, Nvidia said the Isaac GR00T synthetic data workflow and Nvidia Cosmos global foundation models will advance the development of generic humanoid robots.

Over the next two decades, the humanoid robotics market is expected to reach $38 billion. To address this significant demand, particularly in the industrial and manufacturing sectors, Nvidia is launching a set of robotics foundation models, data pipelines and simulation frameworks to accelerate development efforts for next-generation humanoid robots.

The Nvidia Isaac GR00T Synthetic Motion Generation Scheme helps developers create very large synthetic motion data to train their robots using imitation learning.

Imitation learning—a subset of machine learning—enables humans to acquire new skills by observing and imitating specialized human performances. Collecting these comprehensive, high-quality datasets in the real world is laborious, time-consuming, and often very expensive.

Implementing the Isaac GR00T scheme for synthetic motion generation allows developers to easily create very large synthetic datasets from a small number of human demonstrations.

Starting with the GR00T-Teleop workflow, users can leverage Apple Vision Pro to capture human actions in a digital twin. These human actions are simulated by the robot in the simulation and recorded to be used as ground truth.

The GR00T-Mimic workflow then duplicates the captured human demonstration onto a larger synthetic motion dataset. Finally, the GR00T-Gen workflow, built on the Nvidia Omniverse and Nvidia Cosmos platforms, significantly expands this dataset through domain randomization and 3D upscaling.

The dataset can then be used as input for a robot policy, which teaches robots how to move and interact with their environment efficiently and safely in NVIDIA Isaac Lab, an open source modular framework for robot learning.

Global enterprise models are narrowing the simulation-reality gap

Nvidia also announced Cosmos at CES, a platform that includes a family of open, pre-trained global foundation models specifically designed to produce physics-aware video and world-class cases for physical AI development. It includes autoregressive and diffusion models in a variety of sizes and input data formats. The models were trained on 18 quadrillion codes, including 2 million hours of autonomous driving, robotics, drone footage, and synthetic data.

In addition to helping create large datasets, Cosmos can reduce the gap between simulation and reality by upscaling images from 3D to real. Combining Omniverse — a developer platform for APIs and microservices for building 3D applications and services — with Cosmos is crucial, as it helps reduce potential hallucinations typically associated with global models by providing critical guarantees through precise, highly controllable simulations. Great.

Expanding the ecosystem

Collectively, Nvidia Isaac GR00T, Omniverse, and Cosmos help physical AI and human innovation take a giant leap forward. Major robotics companies have begun adopting and showing results with the Isaac GR00T, including Boston Dynamics and Figer.

Manufacturers of robotics software, hardware, and robotics can apply for early access to Nvidia’s Humanoid Robotics Developer Program.

Daily insights into business use cases with VB Daily

If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from organizational transformations to hands-on deployments, so you can share insights to maximize ROI.

Read our privacy policy

Thanks for subscribing. Check more VB newsletters here.

An error occurred.