LG Electronics is accelerating its robotics collaboration with NVIDIA, combining decades of manufacturing and logistics data with NVIDIA’s physical AI technologies to develop a large-scale robot learning ecosystem and strengthen its push into humanoid, industrial and home robotics.
The collaboration centres on LG’s new Data Factory at its Yangjae R&D Campus in Seoul, where robots will collect physical training data across simulated home, manufacturing and logistics environments.
Senior NVIDIA officials recently visited the facility to review the partnership and explore further areas of cooperation. The meeting came four days after LG Group and NVIDIA signed a memorandum of understanding at NVIDIA’s headquarters in California on August 13, signalling an effort to move quickly towards commercial applications.
The Yangjae Data Factory is scheduled to become fully operational by the end of 2026.
Building a data flywheel for robot learning
LG is positioning large-scale, high-quality training data as a core component of its physical AI strategy.
The company plans to combine data accumulated through decades of manufacturing and logistics operations with NVIDIA’s robotics development stack. The objective is to create what LG describes as a “data flywheel”, where robots continuously generate data that can be used to train, refine and improve subsequent generations of robotic systems.
At the Data Factory, LG is deploying its CLOiD home robots across specialised training environments designed to reproduce real-world tasks.
A replicated home allows robots to learn and repeatedly perform cleaning activities, while a simulated manufacturing environment modelled on LG’s washing machine factory in Tennessee trains robots to move, stack and assemble components.
Robots are also being deployed alongside LG CNS logistics automation solutions and in an environment where LG Innotek is training robotic hands.
Data generated across these environments will be connected with NVIDIA’s robotics technologies, allowing physical data to be augmented and supplemented with synthetic training data.
NVIDIA technology supports data-to-deployment pipeline
The collaboration will incorporate NVIDIA Omniverse libraries, Cosmos open world models and the Isaac robotics development platform across LG’s robot development process.
This gives LG tools for simulation, synthetic data generation, robot training and deployment while allowing the company to combine NVIDIA’s physical AI ecosystem with proprietary operational data collected from its own businesses.
LG intends for the Data Factory to become more than a dedicated robot-training facility. It will also serve as a central environment for processing and expanding robot-learning data collected from manufacturing and logistics operations and potentially from home appliances deployed globally.
The four-storey facility spans approximately 10,000 square metres and is expected to accommodate several hundred robots by the end of the year.
LG expects directly collected training data and synthetic data generated and augmented using NVIDIA Cosmos to reach a combined 100,000 hours by year-end, equivalent to approximately 12 years of continuous data.
LG targets humanoid robot development
The resulting datasets will be used to advance LG’s Robot Foundation Model, or RFM, which the company sees as an important foundation for improving humanoid robot performance.
Foundation models for robotics are intended to give machines broader capabilities across multiple tasks rather than requiring separate programming for individual actions. Developing such systems requires large quantities of diverse physical-world data, making the scale and quality of training datasets increasingly important.
LG believes its existing manufacturing footprint provides an advantage because it can generate robot training data from established industrial processes rather than relying entirely on artificial environments.
The company is seeking to combine that operational experience with synthetic data generation and simulation to accelerate learning while continuously feeding improvements back into deployed robotic systems.
Robotics becomes a strategic business for LG
LG has designated 2026 as the starting point for a broader push into robotics and established a Robotics Business Center last month to coordinate its activities across the company.
Reporting directly to the CEO, the centre is responsible for overseeing LG’s robotics businesses and improving execution across different parts of the organisation.
LG already operates in industrial and commercial robotics and plans to expand further into home robotics. Its capabilities also extend to core robotic components such as actuators, giving the company exposure across both hardware and software development.
The Data Factory is intended to add another layer by providing the large-scale data infrastructure required to train and operate increasingly autonomous robotic systems.
“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,” said LG Electronics CEO Lyu Jae-cheol.
Source: MEA Tech Watch Press Reporter
