NVIDIA has introduced Isaac GR00T N1, an open-source foundational model poised to revolutionize humanoid robotics development. This innovative model draws inspiration from human cognition, paving the way for more adaptable and versatile robots.
GR00T N1 incorporates a dual-system approach, mimicking the interplay between fast, reflexive actions and slower, deliberate thought processes in humans. One system handles rapid responses, akin to human reflexes, while the other manages higher-level reasoning, environmental assessment, and action planning. This allows robots to analyze their surroundings and translate complex commands into precise movements. The model’s ability to generalize further enhances its capabilities, enabling it to apply knowledge gained from manipulating one object to interact effectively with others.
Available globally, GR00T N1 empowers robotics developers to customize and build upon the model to address specific needs. A humanoid robot developed by 1X Technologies, trained using GR00T N1, showcases the model’s potential by autonomously performing household tidying tasks.
“The future of humanoids hinges on adaptability and learning,” explains Bernt Børnich, CEO of 1X Technologies. “While we develop our own models, NVIDIA’s GR00T N1 significantly enhances robot reasoning and skills. With minimal post-training data, we fully deployed on NEO Gamma, advancing our mission to create robots that are not just tools, but true companions capable of meaningful human assistance.”
Industry leaders, including Boston Dynamics and Agility Robotics, have already embraced GR00T N1 for their development initiatives. NVIDIA further indicated its commitment to expanding this field by developing a family of related pre-trained robotics models.
“The age of generalist robotics is here,” declares Jensen Huang, NVIDIA’s founder and CEO. “With NVIDIA Isaac GR00T N1 and new data-generation and robot-learning frameworks, robotics developers worldwide will open the next frontier in the age of AI.” This signifies a transformative step towards realizing the potential of general-purpose robots.
