LG, Nvidia Step Up Robotics Push as Madison Huang Leaves 'AMAZNIG LG' Message

LG Electronics and Nvidia accelerate robotics collaboration

Madison Huang visits Seoul data factory to discuss commercialization plans

LG Electronics is accelerating its push into robotics with Nvidia. The company plans to deploy several hundred of its in-house LG CLOiD robots at a robot data factory under construction in Seoul's Yangjae area and use Nvidia's physical AI technology to secure 100,000 hours of training data by year-end, equivalent to roughly 12 years.

LG Electronics said Aug. 18 that it invited key Nvidia officials to the LG Data Factory in Seoul's Seocho district to discuss the direction of their robotics collaboration and commercialization plans. Attendees included LG Electronics Chief Executive Officer Lyu Jae-cheol, LG CNS Chief Executive Officer Hyun Shin-gyoon and LG Sciencepark President Chung Soo-heon, along with other senior executives from the group.

The meeting came four days after the two companies formally announced their robotics partnership in the US. The companies signed a memorandum of understanding on Aug. 13 at Nvidia's headquarters in Santa Clara, California, to cooperate on future strategic businesses. LG Electronics said the latest meeting was meant to move beyond a broad agreement and speed up commercialization.

At the center of the partnership is the Yangjae data factory. LG Electronics is using the site to generate, collect and train data with LG CLOiD robots that it developed and verified in-house. In a home-like space, the robots are trained to clean. In a manufacturing environment modeled on LG's washing machine plant in Tennessee, they learn to transport, load and assemble parts. Robots have also been deployed in areas for LG CNS's logistics automation solution and LG Innotek's robot-hand training.

LG Electronics is focusing on combining its manufacturing data with Nvidia's robotics solutions. The company plans to use data accumulated over decades at manufacturing and logistics sites, along with the know-how of skilled workers, as training material for robots. It also aims to augment and synthesize that data with Nvidia technology to expand the training dataset. LG Electronics sees a data virtuous cycle, in which robots repeatedly train on high-quality data to improve performance, as a core competitive advantage.

The effort will use Nvidia Omniverse libraries, the Cosmos OpenWorld model and the Isaac open robotics development platform. The data factory will span four floors, from one basement level to the third floor above ground, with a total floor area of 10,000 square meters. LG Electronics plans to increase the number of robots deployed there to several hundred by the end of this year.

By adding amplified and synthesized virtual data to data collected directly on site, the company expects its training dataset to reach 100,000 hours by year-end. LG Electronics plans to use the dataset to advance its Robot Foundation Model, or RFM, which it says is critical to humanoid robot performance.

LG Electronics has designated this year as the launch point for a broader expansion of its robotics business. Last month, it set up a Robotics Business Center directly under the CEO. The company aims to expand its robot lineup from industrial and commercial uses into the home and combine key components such as actuators with its data capabilities to become a total robotics solutions provider spanning hardware and software.

Madison Huang, a senior director at Nvidia and the eldest daughter of Chief Executive Officer Jensen Huang, wrote "AMAZNIG LG" on an LG CLOiD robot during her visit to the LG Data Factory. Lyu said LG would strengthen its competitiveness in physical AI and become a total robotics solutions provider through "One LG" synergies across the group and strategic partnerships with global companies.

Kim Dae-young, Hankyung.com reporter kdy@hankyung.com