Project GR00T: NVIDIA’s Humanoid Robots

No, this isn’t Terminator, and this article isn’t science fiction.

This March, tech giant NVIDIA announced Project GR00T, a foundation model for humanoid robots. This groundbreaking initiative aims to drive significant advancements in robotics and embodied AI. NVIDIA also unveiled the Jetson Thor, a new computer for humanoid robots that would use the NVIDIA Thor system-on-a-chip (SoC). The announcement highlighted substantial enhancements to the NVIDIA Isaac robotics platform, incorporating generative AI foundation models and innovative simulation and AI workflow infrastructure tools. These three announcements all point in the direction of powerful robots that would be able to perform economically valuable tasks.

Project GR00T, or Generalist Robot 00 Technology, would empower humanoid robots to comprehend natural language and mimic human actions. These robots are expected to rapidly acquire skills in coordination and dexterity to navigate their environment. Demonstrations showcased robots performing various tasks, underlining the potential for this technology to revolutionize labour.

Jensen Huang is becoming more and more well-known as a charismatic AI leader. Source: CNET

The Jetson Thor computer is tailored for humanoid robots and designed to execute complex tasks. Its architecture is tuned for optimal performance and compactness. The SoC features an impressive GPU based on the NVIDIA Blackwell architecture, a transformer engine capable of 800 teraflops of 8-bit floating point AI performance, and an integrated functional safety processor. This combination of technologies significantly simplifies design and integration processes, making it a cornerstone for future humanoid robotics.

Furthermore, NVIDIA announced significant updates to its Isaac platform, which would be essential for developing new foundation models for any robot in any setting. The updates include Isaac Lab for reinforcement learning and OSMO, a compute orchestration service that facilitates the creation and training of embodied AI models. The new Isaac Lab is a GPU-accelerated application designed for running parallel simulations, which is crucial for robot learning.

The Isaac Manipulator and Isaac Perceptor, new additions to the platform, provide advanced capabilities for robotic arms and multi-camera, 3D surround-vision systems. It’s not hard to imagine a world where AI technology begins to outperform human manufacturing capabilities. This highlights the importance of debate over the uses of AI in the real world. The rate of regulation and policy-making must be able to keep up with the stunning rate of technological development.

Readers should leave with questions about how and to what ends such humanoid robots can be used.

Read More:
Hollow Knight: Silksong, Where Is It Today?
Dune Part Two: Paul’s Prescience – Math Or Magic?