The role of Nvidia in the next generation of human robots

One of the phrases that often repeated in NVIDIA GTC this year was the “physical AI”, a type of artificial intelligence that allows robots to actively work in the material world.

To get there, NVIDIA ISAAC GR00T N1 offers a pre -trained foundation with some basics and is divided into a fast system to take action and a slow planning system.

To teach robots new tasks, a lot of diverse data is needed. The remote shows and the video are important, but it is usually not sufficient. The artificial data created from training in simulation helps fill the gap, and this is where Nvidia’s Omniverse comes with Cosmos.

“Omniverse is a platform that combines data between different sources and allows you to build a physically accurate digital twin,” says Akhil Docca, Omniver’s Project Marketing Manager. “Cosmos is important because it is not only allowed to increase data for light realism, but also allows you to create large (quantity) data.”

This was clarified during the Nvidia Jensen Huang CEO when a digital twin of robot was shown in bright colors that resemble desserts in Omniverse. The same image was shown to ride bicycles through many different repetitions with a vibrant shading and texture added by the universe using simple text claims.

To make these artificial data as necessary on the real world as possible, a detailed representation of physics laws is required. NVIDIA has made a partnership with DeepMind and Disney to develop a Newton’s open -source physics engine. During the keyword, Huang directed Droid BDX, which is controlled by a distance to celebrate this advertisement.

Much of this work is framed on human robots as an answer to the anxiety that is waving on the horizon about the lack of labor. In his main speech, Huang expected that the deficiency around the world would be 50 million workers by the end of the contract and depict in the future as we might pay “$ 50,000 annually” robots to work.

For more clarity about Huang’s comments, I called NVIDIA, which told me that the expected shortfall of 50 million workers by the end of the contract was based on a set of estimates from various sources, industries and sites. The sources included NVIDIA estimates regarding manufacturingand Renewable energyand Truckand nursingHospitality workers in US and EuropeAnd home workers in we and EuropeAnd construction workers in we and Europe.

A representative from NVIDIA also explained the expected pricing model for the company for human robots, saying: “It is more like participating in the full self -driving services of an independent car. The customer buys the car and then pays the services you want in the foreground. We believe that robots will be the same. Buying the robot and pay the services on top.”

To watch these techniques at work, check the video in this article.