Physical AI

Until now, AI has mostly dealt with text and pictures. Physical AI sees its surroundings with cameras and sensors, decides what to do, and moves robot arms or wheels directly. Robots and self-driving cars are the main examples.

The real world is much harder than a screen. Objects slip, people suddenly step in, and a mistake can break something. So a common method is to practice many times in a virtual space inside a computer, then move what was learned to a real robot.

It is the difference between memorizing a cookbook and actually picking up a knife and cooking.

It is seen as the next stage, when AI comes out into factories, warehouses, roads and homes. Nvidia and others used the term widely, and it came to mean the whole robot industry.