NVIDIA Isaac GR00T N1.7: Open Reasoning VLA Model for Humanoid Robots

NVIDIA has introduced GR00T N1.7, an open-source Vision-Language-Action model designed for humanoid robots, emphasizing the importance of human data in enhancing robotic intelligence. The model, which boasts over 20 billion parameters, utilizes a dual-system architecture for high-level reasoning and precise motor control, enabling it to perform complex tasks reliably. With training on more than 20,000 hours of human egocentric video, GR00T N1.7 sets a new standard for dexterity in robotic manipulation, making it ready for commercial deployment in various sectors such as manufacturing and healthcare.