NVIDIA Open-Sources Physical AI Tools for Robotics, Cars

NVIDIA has released a large collection of open-source "physical AI" skills and tools. These are designed to simplify the complex tasks involved in robotics, self-driving cars, vision AI, and industrial digital twins. The new tools are part of the NVIDIA Agent Toolkit. They allow AI agents to directly access NVIDIA’s libraries, models, and frameworks. This helps speed up the development process, from generating data and running simulations to training models and deploying them. Jensen Huang, NVIDIA's CEO, stated that AI agents are revolutionizing software development, and this shift is now coming to physical AI. NVIDIA is redesigning its entire physical AI stack so that agents can directly use its tools. This includes Cosmos for world foundation models, Omniverse for simulation, Isaac for robotics, Metropolis for vision AI, Alpamayo for autonomous driving, and the Jetson platform for edge AI. New skills within the toolkit will help developers turn physical AI development into repeatable instructions for coding agents. There's also a safety component, with the NVIDIA NemoClaw blueprint and OpenShell runtime offering policy-based security and privacy. These new tools aim to accelerate development in areas like robotics, edge AI, and autonomous vehicles. This matters because it could lead to faster advancements in these critical technologies.