NVIDIA has announced the Isaac GR00T N1, described as the world’s first open, fully customizable foundation model designed for humanoid reasoning and skills, during the GTC event. This model aims to aid development in robotics amidst ongoing global labor shortages, estimated to surpass 50 million people.

The GR00T N1 features a dual-system architecture inspired by human cognition, comprising a fast-thinking action model and a slow-thinking decision-making model. It is designed to generalize tasks such as grasping and moving objects and can perform multi-step tasks requiring extensive context.

In collaboration with Google DeepMind and Disney Research, NVIDIA introduced the Newton physics engine, which is being developed to help robots manage complex tasks accurately. This collaboration also includes the MuJoCo-Warp project, which is expected to boost machine learning workloads significantly.

The GR00T N1 foundation model will allow developers and researchers to post-train the system with real or synthetic data for tailored applications. NVIDIA has also unveiled an interactive demo of the GR00T Blueprint for synthetic manipulation, which facilitates the generation of large volumes of synthetic motion data to enhance robot training.

Early access to the GR00T N1 is granted to various humanoid developers, including Agility Robotics and Boston Dynamics, who are expected to leverage the model's capabilities for different applications in the robotics sector. The training data and evaluation scenarios for GR00T N1 are accessible via platforms like Hugging Face and GitHub.