Nvidia has unveiled Groot N1, an advanced AI model for humanoid robotics, during its GTC 2025 conference. Drawing inspiration from human cognitive processes, Groot N1 features a dual-system design that enables "fast and slow thinking," allowing robots to interact with their environment more effectively.

This model represents the next step in Nvidia’s robotics journey, evolving from Project GR00T, which was introduced at GTC 2024. While GR00T targeted industrial use cases, Groot N1 broadens its application to encompass diverse robotic form factors, paving the way for wider adoption.

The "slow thinking" component empowers robots to interpret their surroundings, process instructions, and deliberate before acting. On the other hand, the "fast thinking" system converts these deliberations into seamless robotic actions, even for tasks requiring multiple steps.

Groot N1 will be available as an open-source platform, providing developers with the tools they need to innovate. Alongside the model, Nvidia will release simulation frameworks and blueprints for synthetic training data generation, further enhancing the ecosystem for robotics development.

During the conference, Nvidia demonstrated Groot N1’s capabilities using Blue, a humanoid robot co-developed with Disney and Google DeepMind. Blue also leverages Newton, a cutting-edge engine that boosts learning speeds by 70 times, showcasing the synergy between Nvidia’s hardware and software innovations. 

Main image: nvidia.com

Comment on this