A frequent repeated phrase in the Nvidia GTC this year is “physical AI,” the type of AI that allows robots to work effectively in the physical world.
To get there, Nvidia offers the Isaac Gr00T N1, an open-source pre-trained foundation model with some of the basics and damaged in a quick system for action and a slow system for action planning.
Display of how teleoperation can be useful -benefit to teaching robots of new skills
For teaching robots of new tasks, many different data are required. Tele-operation and video demonstrations are important but are usually not enough. Synthetic data generated from simulation training helps fill in the gap, and here's where Nvidia's Omniverse enters.
“Omniverse is a platform that combines data between different sources and gives you to produce a digital twin accurate physical,” said Akhil Docca, Senior Project Marketing Manager for Omniverse. “Cosmos are important because not only is it adding data for photorealism, it allows you to create an exponentially large [amount] of data. “
It was illustrated during the Nvidia CEO keynote speech Jensen Huang when a digital twin of a robot was shown in bright candy colors in Omniverse. The same image is shown in cycling through many different iterations with living shading and textures added by the cosmos using simple text signals.
Nvidia CEO Jensen Huang shows the company's Omniverse with Cosmos
To make the synthetic data appropriate to the real world as much as possible, a detailed representation of the laws of physics is required. NVIDIA has cooperated with Deepmind and Disney to develop open-source Newton physics. During his keynote, Huang released a remote-control BDX Droid to celebrate the announcement.
Nvidia CEO Jensen Huang with a BDX droid from Disney Star Wars
Most of this work on the humanoid robots is framing as an answer to anxiety outing anxiety about labor shortages. In his main tone, Huang predicted a worldwide shortage of 50 million workers by the end of the decade and thought of a future where we could pay “$ 50,000 per year” robots to work.
To get more clarity on Huang's comments, I contacted Nvidia, who told me the predicted deficiency of 50 million workers by the end of the decade was based on a combination of estimates from various sources, industries and locations. Resources provided by Nvidia include estimates about Working, Changes energy, Trucking, Take careHappiness workers at United States and Europedomestic workers at We are and Europeand construction workers in We are and Europe.
A representative from Nvidia also clarified the company's expected model of pricing for humanoid robots, saying “This is more like subscribing to the entire driving service for an autonomic vehicle. The customer buys the car and then pays for the services it wants above. We believe the robots will be the same. Buy a robot and pay for the services on top.”
To see these technologies acting, check out the video in this article.
