Madison Huang (right), senior director of Omniverse and Robotics at Nvidia, smiles as she receives a bouquet from “LG CLOi” at the LG Data Factory being built at LG Electronics’ Yangjae R&D Campus in Seocho-gu, Seoul, on the 18th. LG Electronics
LG Electronics (066570.KS) CEO William Cho met with Madison Huang, senior director of Omniverse and robotics at Nvidia and the eldest daughter of Nvidia CEO Jensen Huang, and agreed to secure roughly 12 years’ worth of robot training data within this year. With senior executives from the two companies meeting again just four days after LG Group Chairman Kwang-mo Koo sat down with Nvidia’s chief, cooperation on physical artificial intelligence is gaining momentum.
LG Electronics said on the 18th that key officials from both companies, including Cho and Huang, met at its Seoul R&D campus in Yangjae to discuss cooperation on robotics and physical AI.
The two companies set a goal of securing more than 100,000 hours of robot training data within this year by applying Nvidia’s physical AI technology to the data factory that LG Electronics is building. That volume equals about 12 years of data, assuming a single robot accumulates data without pause.
The data factory is a kind of “robot training center” where robots repeatedly perform actual tasks to accumulate data, which is then verified and refined. It is being built with one basement level and three above-ground floors, spanning a total floor area of 10,000 square meters. LG Electronics plans to deploy hundreds of robots within this year, including its humanoid “LG CLOi.” Nvidia will apply its Omniverse libraries, Cosmos open-world models and Isaac robotics platform.
After touring the site for about two hours that day, Huang praised it as “incredible.” Cho said, “We will secure competitiveness in physical AI by combining the capabilities of One LG with collaboration with global partners.”
In the industry, securing high-quality data is regarded as the key factor determining competitiveness in physical AI. Unlike text- and image-centered generative AI, robots must learn vast amounts of complex real-world data, such as the movements and work know-how of on-site workers. LG Electronics plans to combine the data it has accumulated in manufacturing and logistics settings with Nvidia’s technology to accelerate development of a robot foundation model (RFM), which serves as the “brain” of a robot.
The partnership will also extend to development of next-generation humanoids. Koo and Huang met at Nvidia’s U.S. headquarters on the 13th and signed a memorandum of understanding (MOU) to develop a next-generation bipedal humanoid next year, with a target of unveiling it in the first quarter of next year. LG Electronics is also pushing to conduct a CLOi demonstration at its plant in Tennessee within this year.
Last month, LG Electronics established a robotics business center reporting directly to the CEO and began mass production of “Axium,” an actuator for robot joints. It aims to grow into a “total robotics solution provider” spanning both robot hardware and software.
Madison Huang (right), senior director of Omniverse and Robotics at Nvidia, signs “LG CLOi” with the phrase “AMAZING LG” at the LG Data Factory being built at LG Electronics’ Yangjae R&D Campus in Seocho-gu, Seoul, on the 18th. LG Electronics
Ryu Jae-cheol (first from right), CEO of LG Electronics, and Madison Huang (second from right), senior director of Omniverse and Robotics at Nvidia, examine a data augmentation and synthesis solution at the LG Data Factory being built at LG Electronics’ Yangjae R&D Campus in Seocho-gu, Seoul, on the 18th. LG Electronics