For the first time, factories can now run robots from competing manufacturers as a unified team without custom engineering for each integration. LG CNS, the IT services division of South Korea's LG Group, unveiled PhysicalWorks on Thursday, a software platform designed to manage mixed-brand robot fleets in real time. This addresses one of the most persistent obstacles in industrial robotics: getting machines from different makers to work together seamlessly.

Why Can't Different Robots Just Work Together?

Until now, coordinating robots from multiple manufacturers required expensive custom engineering work tailored to each specific combination of machines. Each robot operates on its own control system, uses different communication protocols, and has unique capabilities. Imagine trying to get four people from different countries to work on the same assembly line without a common language or shared instructions. That's the problem factories have faced for decades.

The challenge has become more urgent as companies increasingly want flexibility in their automation strategies. Rather than committing to a single robot vendor, manufacturers prefer to choose the best tool for each specific task, whether that's a bipedal humanoid for picking tasks, a wheeled quadruped for transport, or a specialized gripper robot for delicate assembly work.

How Does PhysicalWorks Actually Work?

PhysicalWorks operates in two complementary modules. The first uses simulation and video data to train robots on specific tasks before they ever touch a real object. The second module assigns and reassigns tasks across mixed-brand robot fleets in real time, dynamically adapting when circumstances change.

To demonstrate the platform's capabilities, LG CNS ran a live demonstration at its Magok campus in Seoul featuring four robots from four different manufacturers working in concert. A bipedal humanoid from China's Unitree picked up a packaged item and placed it in a box. A wheeled-leg quadruped from Deep Robotics then took the box and carried it to Dexmate's wheeled humanoid, which loaded it onto an assigned shelf. The quadruped returned for the next cycle while the wheeled humanoid placed an empty box back on the conveyor for the bipedal robot to refill. When a staged emergency reassigned the quadruped to patrol duty, the platform automatically pulled in a logistics robot from Bear Robotics to take over the transport task.

None of the robots required remote control, and the handoff between two of them, roughly two to three meters apart, took about 90 seconds. A company representative noted that the robots would "move faster as they accumulate more field training".

What Makes LG CNS Uniquely Positioned for This?

LG CNS brings four decades of experience building factory IT systems for Korean manufacturers like LG Electronics and LG Display. This deep knowledge of legacy production software and manufacturing workflows is what makes integration possible at scale, according to company leadership.

"We source whichever robots fit the site, train them on the client's data, and run their workloads through our platforms," said CEO Hyun Shin-gyoon, explaining that the company's grasp of legacy production software is what makes integration possible at scale.

Hyun Shin-gyoon, CEO at LG CNS

The company has been building toward this launch for 11 months. LG CNS invested in US robot-brain startup Skild AI last June, took a stake in Dexmate in March, and opened a robotics consulting unit in April. The company went public on South Korea's Kospi exchange in February 2025.

Steps to Understanding Robot Fleet Management in Modern Factories

• Task Assignment: The platform continuously monitors which robots are available and what tasks need completion, automatically routing work to the most appropriate machine in real time.
• Data Training: Each robot learns from simulation and video data specific to the client's facility before deployment, reducing errors and improving efficiency.
• Dynamic Reallocation: When circumstances change, such as equipment failure or emergency situations, the system instantly reassigns tasks to available robots without human intervention.
• Cross-Manufacturer Compatibility: The platform translates between different robot control systems and communication protocols, eliminating the need for custom engineering on each integration.

Where Is This Technology Being Tested?

LG CNS is running more than 20 proof-of-concept projects with clients across electronics, chemicals, batteries, and shipbuilding, including welding work at shipyards facing labor shortages. The fleet-management module has been operating since December at Busan's national pilot smart city, coordinating patrol, barista, luggage, and cleaning robots.

When asked about the timeline for revenue generation, Smart Logistics and City division head Lee Joon-ho indicated that meaningful results were "around one to two years" away. LG CNS posted 6.13 trillion won, approximately $4.22 billion, in revenue last year.

What Does This Mean for the Broader Robotics Industry?

PhysicalWorks represents a significant shift in how industrial robotics will be deployed going forward. Rather than factories being locked into single-vendor ecosystems, they can now mix and match robots based on specific needs and cost considerations. This approach mirrors how software companies have built platforms that work across different hardware manufacturers, bringing flexibility and competition to the robotics market.

The timing is significant. As humanoid robots and advanced mobile robots continue to improve, factories need orchestration platforms that can coordinate these diverse machines. LG CNS's entry into this space, backed by its manufacturing software expertise and strategic investments in robot-brain technology, signals that the industry is moving beyond individual robot deployments toward integrated, multi-robot systems that can adapt to changing production demands in real time.