Tech Desk
Aug 06, 2024
It's rarely heard that a tech company is making use of a product manufactured by its rival since this would portray the impression that it does have some inclination towards the opposing party's developments.
Much is the case with Apple and NVIDIA, although both compete in different domains of the tech industry, there seems to be a rivalry when it comes to gaining ground in the sphere.
In a surprising development, NVIDIA is said to be using Apple Vision Pro to make its humanoid robots better understand visual dynamics.
Read more: iOS 18 beta 5 to launch today, add new ‘Distraction Control’ feature to Safari
The development emerged from a video NVIDIA shared on YouTube, demonstrating how deep in production its humanoid robots have gone.
Citing an NVIDIA press release, 9to5Mac states that training the basic AI models for humanoid robots requires a tremendous amount of data.
Given that, one way of providing the AI models with footage of human demonstrations is teleportation, but this method costs companies exorbitantly and takes time.
In this process, a graphical demonstration of an NVIDIA AI- and Omniverse-enabled teleoperation reference workflow allows researchers and AI developers to produce huge amounts of data based on synthetic motion and perceptions from the least possible demonstration of human workflow.
To attain this feat, NVIDIA developers first used Apple Vision Pro to record very few teleoperated demonstrations. Then they replicate those demonstrations in NVIDIA Isaac Sim and utilise the MimicGen NIM microservice to produce synthetic datasets from recorded human workflows.
"The developers train the Project GR00T humanoid foundation model with real and synthetic data, enabling developers to save time and reduce costs. They then use the Robocasa NIM microservice in Isaac Lab, a framework for robot learning, to generate experiences to retrain the robot model. Throughout the workflow, NVIDIA OSMO seamlessly assigns computing jobs to different resources, saving the developers weeks of administrative tasks," stated the NVIDIA press release.
