South Korea’s LG Electronics, said on Aug. 18 that it plans to use Nvidia software to train robots with 100,000 hours of real and synthetic data by the end of the year at its 10,000-square-meter Yangjae Data Factory, which is under construction in Seoul.

The move follows an August 13 memorandum of understanding signed by LG Group Chairman Koo Kwang-mo and Nvidia CEO Jensen Huang that covers robotics, AI factories, and mobility.

The companies have also set up a joint task force.

LG said the site will use Omniverse, Isaac, and Cosmos. Omniverse and Isaac will support robot simulation, testing, and synthetic data generation, while Cosmos will add to virtual datasets.

LG said it expects to deploy hundreds of CLOiD humanoid robots at the facility by the end of the year.

The robots will learn home-cleaning tasks, as well as transport, loading, and assembly work, in a simulated factory environment modeled on LG’s washing machine plant in Tennessee.

Separately, LG said its wheel-based CLOi robot will begin factory verification in Tennessee this year, while Nvidia said Boston Dynamics, Figure, and Germany-based Neura Robotics are building humanoid robots on the same software stack.

🔗 Source: Chosun Daily