Nvidia’s role in the next generation of humanoid robots

A often recurring phrase of this year’s NVIDIA GTC was “physical AI”, the species of AI that allows robots to work effectively in the physical world.

To get there, NVIDIA offers its ISAAC GR00T N1, an open source foundation model that has been pre -trained with some of the basics and is broken into a quick action system and a slow action planning system.

A variety of data is needed to teach robots new tasks. Teleoperative and video demonstrations are important, but are usually not enough. Synthetic data generated by training in simulation help to fill the gap and here comes the NVIDIA COSMOS.

“Omniverse is a platform that combines data between different sources and allows you to build a digital twin that is physically accurate,” says Achilles Docca, a senior marketing manager for Omniverse projects. “Cosmos is important as it not only allows for an increase in photorealism data, but also allows you to create an exponentially large (quantity) data.”

This was illustrated during the main speech of NVIDIA CEO Jenson Huang, when a robot’s digital twin was shown in bright colors that resemble candy. The same image was shown in cycling through several different iterations with vital shading and texture added by Cosmos, using simple text prompts.

In order to make these synthetic data as possible to the real world as possible, a detailed presentation of the laws of physics is required. NVIDIA has partnered with Deepmind and Disney to develop an open -code physics engine. During his main note, Huang removed a remotely controlled BDX Droid to celebrate the message.

Much of this work on humanoid robots has been defined in response to anxiety from labor shortages. In his main note, Huang predicts a shortage of 50 million workers by the end of the decade and foresees a future where we can pay $ 50,000 a year robots for work.

To get more clarity about Huang’s comments, I contacted NVIDIA, who told me that the estimated shortage of 50 million workers by the end of the decade was based on a combination of estimates from various sources, industries and places. The sources provided that NVIDIA include forecasts regarding production., renewable energy., freight., breastfeedingHospitality workers in USA and Europedomestic workers in USA and Europeand construction workers in USA and EuropeS

A representative of NVIDIA also clarified the expected model of the pricing company for humanoid robots, saying: “Rather, it is like a subscription for full services for self-driving autonomous vehicle. The customer buys the car and then pays for the services it wants from above. We believe that the robots will be the same.”

To see these technologies in action, check out the video in this article.