With Project GR00T, their new foundation model, Nvidia is making a foray into the humanoid robotics space.
The company announced the artificial intelligence (AI) system at its developer conference this week, saying that it is versatile enough to handle a multitude of tasks, including generating text to produce videos and images.
Through observation and learning from human behavior, Project GR00T (Generalist Robot 00 Technology) aims to equip humanoid robots with the ability to comprehend natural language and mimic human movements. This makes it possible for robots to quickly pick up the abilities—like dexterity and coordination—necessary for navigating, adjusting to, and interacting with the human world.
“One of the most exciting problems to solve in AI today is building foundation models for general humanoid robots,” Nvidia founder and CEO Jensen Huang stated at the conference. Leading roboticists worldwide are making enormous strides towards artificial general robotics thanks to the convergence of enabling technologies.
Apart from Project GR00T, Nvidia has added two more features to its Isaac robotics platform: the Isaac Perceptor, which provides robots with enhanced visual capabilities, and the Isaac Manipulator, which is a set of foundation models for controlling robotic arms.
Moreover, Nvidia unveiled the Jetson Thor computer, a humanoid robot’s central processing unit that helps with difficult jobs and guarantees secure and seamless interactions between humans and machines.
This week, PYMNTS looked at Nvidia’s B200 chip, which experts believe has the potential to revolutionize the e-commerce sector by enabling AI applications to operate more quickly and effectively.
The new chips may eventually lead to lower prices and wider availability for AI products. This is because of “cheaper inferencing,” according to Serene Data Ops founder Abdullah Ahmed, who spoke with PYMNTS about it.
Based on chips such as B200, more businesses may now start utilizing basic AI models that are sufficient for their purposes, according to Ahmed. They can then focus on enhancing and marketing their products.
This means that even smaller companies might be able to turn to AI products without needing as many costly GPUs, which in turn means that developing and using AI tools could cost less, making them more affordable for businesses and consumers.
But in light of the tremendous demand, he added, “this depends on Nvidia’s ability to maintain reasonable lead times.” “The B200 is completely focused on the massive language model craze, like GPT4 from OpenAI, which powers ChatGPT.”