Jensen Huang, CEO of Nvidia, says, “Building foundation models for general humanoid robots is one of the most exciting problems to solve in AI today.” This is hard to disagree with. The humanoid form factor is one of the most controversial topics in robots right now. It’s bringing in a lot of venture capital while also making a lot of people very skeptical.
Nvidia wants a piece, of course. In recent years, the chip giant has made a strong case for itself as a leader in robotics innovation through projects like Isaac and Jetson. It is now probably the most important hardware company in AI. The company is putting down its humanoid flag this week at its annual GTC developer meeting. Project GR00T may or may not be a reference to Marvel’s talking space tree from the movie Iron Man.
The company that makes chips calls the new platform “a general-purpose foundation model for humanoid robots.” A lot of new companies have started working in this area recently, such as 1X Technologies, Agility Robotics, Apptronik, Boston Dynamics, Figure AI, Fourier Intelligence, Sanctuary AI, Unitree Robotics, and XPENG Robotics. This company is basically making an AI platform for all of them. At the moment, that includes almost all of the major companies that make humanoid robots, with a few big exceptions like Tesla.
A quote from co-founder and Chief Robotics Officer Jonathan Hurst gives Agility more attention in the announcement: “We are at an inflection point in history, with human-centric robots like Digit poised to change labor forever.” Modern AI will speed up progress, making it possible for robots like Digit to assist people in all parts of their daily lives. We’re excited to work with NVIDIA to set up the computers, simulation tools, machine learning environments, and other structures that will make the dream of robots being a part of everyday life a reality.
Geordie Rose, co-founder and CEO of Sanctuary AI, also has something to say: “Embodied AI will not only help solve some of the world’s biggest problems, but it will also lead to innovations that are currently beyond our reach or imagination.” This important of technology shouldn’t be built separately, which is why we put long-term partners like NVIDIA first.
GR00T will also work with new gear from Nvidia. Jetson Thor is a new computer made just for running simulation processes, generative AI models, and more in a humanoid form factor. It keeps things in the Marvel Cinematic Universe going. I still warn people not to use vague terms like “general purpose” to describe these machines, but Nvidia’s strong interest is good news for the category and will almost certainly speed up development.
Nvidia Says This About The New Chip:
The SoC has a new GPU built on the NVIDIA Blackwell architecture and a transformer engine that gives it 800 teraflops of 8-bit floating point AI performance. This is enough to run multimodal generative AI models like GR00T. It makes design and integration a lot easier because it has a functional safety processor built in, a high-performance CPU cluster, and 100GB of Ethernet speed.
Even though general purpose won’t be ready for years, letting third-party coders use it will help a lot in closing the gap.
Two more important robotics projects, Isaac Manipulator and Isaac Perceptor, were announced at GTC this week. For many years now, manipulation has been an important part of robots. Leading the way were the huge industrial computer arms that are now commonplace in making cars. People in the next age will be even faster and more flexible. It makes sense that Nvidia wants in on the action.
“Isaac Manipulator offers cutting-edge dexterity and modular AI capabilities for robotic arms, with a strong set of foundation models and GPU-accelerated libraries,” the business says. The zero shot perception improves efficiency and throughput, making it possible for developers to automate more new robotic jobs. It speeds up path planning by up to 80 times.
Some big names have already joined Nvidia. These are Solomon, Universal Robots, Yaskawa, Franka Robotics, and PickNik Robotics.
Robots that can move around on their own are also getting some love in the form of Perceptor. Nvidia has long been interested in vision processing for robots, and this study continues that focus. This is aimed directly at “multi-camera, 3D surround-vision capabilities.” Already, ArcBest, BYD, and KION Group have signed up.
Also Read: Reddit’s Planned Ipo Share Price Looks High Until You See How Much Money It Makes From Ai
The race for market share between humanoids and mobile manipulators will be very interesting over the next few years, and Nvidia wants to be a part of it all.
What do you say about this story? Visit Parhlo World For more.
