Overview
- NVIDIA senior director Madison Huang visited LG's Yangjae Data Factory on Tuesday to inspect progress and confirm integration of NVIDIA's robotics software with LG's deployed CLOiD units.
- The Yangjae facility is a four-floor, 10,000 square meter site that LG says will house several hundred robots by year-end and serve as a centralized hub for task data collection.
- LG has set a target to collect roughly 100,000 hours of real and synthetic robot training data at Yangjae by the end of the year to advance its Robot Foundation Model.
- Data gathered by CLOiD units in replicated home and factory spaces will be augmented and synthesized using NVIDIA Omniverse libraries, Cosmos open‑world models and the Isaac robotics platform before feeding model training.
- LG has created a Robotics Business Center reporting to the CEO and says the data pipeline supports a roadmap that includes pilot production sites, a planned AI factory and a next‑generation bipedal humanoid slated for unveiling in Q1 2027.