Nvidia introduced technologies to enhance humanoid robot development, including Isaac GR00T N1, a customizable foundation model for humanoid reasoning and skills.
The technologies include simulation frameworks, such as the Isaac GR00T Blueprint for synthetic data generation, and Newton, a physics engine developed with Google DeepMind and Disney Research.
GR00T N1 accelerates industries facing labor shortages by offering pretrainable models for robotics developers worldwide.
Nvidia's CEO highlighted that generalist robotics are emerging with the new technologies and data-generation frameworks.
The GR00T N1 model features a dual-system architecture inspired by human cognition, enabling fast-thinking and slow-thinking robotic decision-making.
Developers can generalize GR00T N1 across various tasks and industries, enhancing robot reasoning and skills.
Huang demonstrated a humanoid robot performing tidying tasks autonomously with post-trained policies built on GR00T N1.
Collaborations with leading humanoid developers worldwide have provided early access to GR00T N1, aiming to advance robot learning capabilities.
Nvidia, Google DeepMind, and Disney Research are developing Newton, an open-source physics engine to improve robot learning precision and task handling.
The Nvidia Isaac GR00T Blueprint aids in generating synthetic motion data for manipulation tasks, significantly reducing the need for extensive human demonstration data.