- LG's new Data Factory in Seoul aims to generate 100,000 hours of robotic training data by 2026
- Collaboration leverages real and synthetic data to accelerate robot development and deployment
- Focus on supporting LG's Robot Foundation Model and expanding robotics business across sectors
LG Electronics and NVIDIA are pushing their robotics collaboration further through LG’s new Data Factory in Seoul, a facility designed to combine robotics data, manufacturing expertise and artificial intelligence to advance physical AI. The center is located at LG’s Yangjae R&D Campus and covers about 10,000sqm. It is expected to house several hundred robots by the end of 2026 and generate 100,000 hours of training data, according to both companies. They reviewed the facility’s development on August 17, four days after signing a memorandum of understanding at NVIDIA’s headquarters in Santa Clara, California, on August 13.
The goal behind the setup is to provide robots with controlled environments where they can repeatedly perform tasks, gather data and refine their capabilities. LG is using its CLOiD home robots in simulated settings that replicate home, manufacturing and logistics environments. One area mirrors a home setting, where robots perform cleaning tasks. Another recreates elements of LG’s washing machine manufacturing operations in Tennessee, allowing robots to move, stack and assemble components. Data is also being generated through logistics automation deployments at LG CNS and robotic-hand training at LG Innotek.
LG plans to connect this data with NVIDIA’s robotics technologies, including NVIDIA Omniverse libraries, Cosmos open world models and the Isaac robotics development platform. The combination is intended to increase the volume and quality of data available for robot training by augmenting and synthesizing physical-world data.
By the end of 2026, LG expects data collected at the facility, together with synthetically generated and augmented data, to reach 100,000 hours. LG equates this volume to roughly 12 years of training data.
This data strategy is central to LG’s development of its Robot Foundation Model (RFM), which is intended to support the capabilities of its humanoid and other robots. The company has identified 2026 as a starting point for expanding its robotics business. In July, LG established a Robotics Business Center reporting directly to its CEO. The unit is responsible for coordinating the company’s robotics activities and improving execution across the business.
LG already operates in industrial and commercial robotics and is expanding its focus toward home robotics. Its capabilities span robotic components, including actuators, as well as complete robotic systems and manufacturing infrastructure.
The Data Factory adds a dedicated infrastructure layer for collecting, validating and scaling the data required to train robots. The collaboration with NVIDIA could allow LG to combine this physical-world data with simulation and synthetic data to accelerate development across different robotic applications.
For robotics developers, access to high-quality physical-world data is becoming an important part of training increasingly capable systems. Unlike conventional AI models that can primarily learn from digital datasets, physical AI systems must learn how to interact with real environments, objects and machines.
LG’s manufacturing and logistics operations provide a large source of operational data that can be used for this purpose. Linking those datasets with NVIDIA’s robotics software and simulation technologies gives the companies a framework for continuously generating, augmenting and validating training data.
“Through the synergy built on ‘One LG’ – bringing together core capabilities across the Group – and strategic collaboration with global partners, we will secure our competitiveness in physical AI and become a comprehensive robotics solutions provider with both hardware and software capabilities,” said Lyu Jae-cheol, CEO of LG Electronics.
The immediate focus is on expanding the Data Factory’s capabilities and building the training datasets needed to advance LG’s robotics models. Over time, the approach could determine how quickly LG can move from robotics experimentation to deployment across manufacturing, logistics and consumer environments.
Takeaways
- - LG and NVIDIA are building a robotics data pipeline centered on real-world and synthetic training data.
- - The Seoul Data Factory is designed to support physical AI by combining manufacturing, logistics and home-environment simulations.
- - LG’s strategy depends on scaling high-quality data for its Robot Foundation Model and broader robotics business.
- - The collaboration highlights how sourcing, electronics, logistics and mobile-style automation ecosystems are converging in the next phase of robotics development.
- - If successful, the model could help shorten the path from robotics research to practical deployment in homes and industrial settings.
Disclaimer: This article may have been created with AI assistance and reviewed by our editorial team. It is provided for general informational purposes only. Readers should verify information independently before relying on this content.
