NVIDIA announced Project GR00T Foundation Model for Humanoid Robots and Major Isaac Robotics platform update

Isaac Robotics Platform Now Provides Developers New Robot Training Simulator, Jetson Thor Robot Computer, Generative AI Foundation Models

NVIDIA has introduced Project GR00T, a versatile foundation model tailored for humanoid robots, as part of its efforts to advance breakthroughs in robotics and embodied AI. In conjunction with this initiative, the company has unveiled a new computing system called Jetson Thor, specifically designed for humanoid robots and based on the NVIDIA Thor system-on-a-chip (SoC). Jetson Thor is engineered to serve as a cutting-edge computing platform capable of executing complex tasks while ensuring safe and natural interactions with humans and machinery. Featuring a modular architecture optimized for performance, power efficiency, and size, Jetson Thor incorporates a next-generation GPU based on the NVIDIA Blackwell architecture, delivering 800 teraflops of 8-bit floating point AI performance to support multimodal generative AI models like GR00T.

NVIDIA's comprehensive AI platform for leading humanoid robot companies includes collaborations with prominent entities such as 1X Technologies, Agility Robotics, Apptronik, Boston Dynamics, Figure AI, Fourier Intelligence, Sanctuary AI, Unitree Robotics, and XPENG Robotics, among others. Embodied AI models necessitate substantial amounts of both real and synthetic data. To address this need, the new Isaac Lab offers GPU-accelerated, lightweight, and performance-optimized simulations tailored for robot learning. Meanwhile, OSMO facilitates the coordination of data generation, model training, and software/hardware-in-the-loop workflows across distributed environments to scale robot development workloads effectively.