Nvidia's new world model helps robots navigate the world
Summary
Nvidia introduced Cosmos 3, a new AI model that helps robots and autonomous vehicles understand and predict what will happen in real-world environments. This open model generates data about how machines move, which developers can use to train robots and other machines to navigate and act better in physical spaces.Key Facts
- Cosmos 3 is trained on 20 trillion pieces of data, including images, videos, sounds, text, and robot movements.
- Unlike regular AI that creates images or videos, Cosmos 3 models machine actions like joint movements and paths.
- Developers can use Cosmos 3 to simulate physical tasks and build specialized models for robots and vehicles.
- The model is open, allowing hardware makers to customize it for their needs.
- Nvidia is working with companies like Agile Robots and Runway to support Cosmos 3.
- Cosmos 3 can create rare or risky scenarios for training that are hard to replicate in real life, such as robot accidents.
- Nvidia offers two versions now: a high-accuracy "super" model and a fast "nano" model, with a local "edge" version coming soon.
- Nvidia aims to make AI that not only answers questions or creates images but can plan and act in the physical world.
Read the Full Article
This is a fact-based summary from The Actual News. Click below to read the complete story directly from the original source.