Generative artificial intelligence will help robots perform actions based on input data - combining text and speech, video, "human demonstrations", and past experience.
As noted by Reuters, components of the generative AI called Project GR00T will complement Nvidia's existing hardware and software platform for robotics.
"Smarter and faster robots will work in heavy industry around the world," says Rev Lebaredian, Vice President of Omniverse and Simulation Technology. "We work with the entire global robot ecosystem to accelerate development and deployment."
"Jetson Thor" is a powerful computer that powers this generative artificial intelligence software and is based on one of Nvidia's AI microchips. According to the company, it will provide robots with enough computational power to perform complex tasks and interact with humans and machines.
The tools include the ability to teach robots to make better decisions over time - a process known as "reinforcement learning." Nvidia also stated that they will release pre-trained models of robotic devices and other software that will enhance the capabilities of robotic arms and multi-camera sensory work.
Just a week ago, the startup Figure AI demonstrated the first results of collaboration with OpenAI - in the video, the Figure 01 robot effortlessly processed requests visually and auditorily, provided full answers or performed actions, responding to human requests.
Comments (0)
There are no comments for now