Nvidia is helping humanoid robots learn through Apple Vision Pro instruction | TechCrunch

Photo of author

By admin


Robotics has been a fundamental component of Nvidia's stratospheric growth in recent years. When the chip maker announced the GR00T March last yearIt was heralded as a watershed moment in humanoid robotics. Most of the division's biggest names were present at the launch, including 1X Technologies, Agility Robotics, Apptronic, Boston Dynamics, Figure AI, Fourier Intelligence, Sanctuary AI and Unity Robotics.

A CES on Monday, CEO Jensen Huang unveiled Another essential aspect of the GR00T: the blueprint. The new method is based on imitation learning, a fundamental aspect of teaching a robot a new skill. This is doubly important in the world of humanoids, whose designs are based on teaching.

Sounds like imitation learning. A person performs an action and the robot follows. It is an efficient learning method designed to automate existing tasks, such as in factories and warehouses, where the first few rounds of humanoid deployment are underway.

Teleoperation has a fundamental role in all this. It can be used to teach robots remotely, instantly digitizing one's actions by approximating a real-life environment.

With the GR00T blueprint, users can create these actions using Apple's Vision Pro. The process is captured as a digital twin, which the robot can run repeatedly in simulation.



Source link

Leave a Comment