LG CNS on the 7th unveiled PhysicalWorks, a robot transformation (RX) platform that shortens the deployment period for industrial robots, which used to take months, to 1 to 2 months. The company said it will use this as a starting point to fully expand its Robotics business, cited as a future growth engine.
LG CNS presented PhysicalWorks, which integrates and manages the full cycle from robot data collection, training, verification, field application, operation, to control, at the RX Media Day held at LG Science Park in Magok, Seoul, that day. LG CNS is the first to introduce an end-to-end RX platform brand that spans from robot training to operation. The key is to quickly train intelligent robots and settle them into the field.
PhysicalWorks consists largely of two core platforms: PhysicalWorks Forge and PhysicalWorks Baton.
PhysicalWorks Forge is designed with the goal of training and hardening robots to a level where they can be deployed in real-world environments. It handles the entire process in one place, from data collection needed for robot training to robot verification and field application. Forge means "to harden."
A hallmark of Forge is its use of simulation-based data training. An LG CNS official said, "Moving beyond the existing approach of having robots mimic human actions thousands of times, we improved efficiency by using simulation data that recreates actual sites and tasks in a 3D virtual environment for training." The company plans to additionally apply methods that convert videos of human work into training data or use motion capture.
The selection process for the collected data is also automated to filter, organize, and process it, improving training efficiency. For example, in the training process where a robot picks up products in a factory, only successful actions are selected, or unnecessary segments are removed.
Robots that have completed training are validated for task feasibility and stability through simulations in a 3D virtual environment, then optimized for the actual site before deployment. The company emphasized, "With this, the time it takes to train a robot and deploy it on site can be reduced from several months to 1 to 2 months."
Baton is a platform that issues tasks to and provides integrated control and monitoring of robots in various forms, including bipedal, quadrupedal, and wheeled types. It supports operating robots of different manufacturers and various forms within a single system.
An LG CNS official said, "Baton standardizes and systematizes robot operating status and control information, enabling manufacturers to manage different types of robots efficiently."
Baton automatically allocates tasks by robot using mathematical optimization and prevents collisions by optimizing travel routes. Agentic AI reflects task progress and equipment status changes in real time to adjust workflow. For example, if a conveyor belt stops, logistics routes are automatically reconfigured, and if a particular robot stops, the task is immediately switched to another robot.
LG CNS projected that applying PhysicalWorks Baton to a robot operating environment of about 100 units, including autonomous mobile robots (AMRs) and automated guided vehicles (AGVs), would increase productivity by about 15% or more and cut operating costs by up to 18%. The company said the effect is particularly pronounced in environments where robots from various manufacturers are mixed, because duplicate movements, congestion, and manual intervention decrease.
According to LG CNS, PhysicalWorks Forge is currently conducting proof-of-concept (PoC) projects with more than 20 customers. PhysicalWorks Baton is being used to provide integrated control for four types of robots—patrol, barista, baggage-carrying, and cleaning—in the Busan Smart City national pilot city project.
Hyun Shin-Gyoon, president of LG CNS, said, "The essence of RX is not determined by merely securing robots, but by whether robots can be quickly settled into the field to work and be operated sustainably even in changing environments."
He added, "Based on full-stack capabilities that span from establishing robot adoption strategies optimized for customer sites to securing industry-specialized robot foundation models (RFMs) and robot training, application, and operation, we will set a new standard for the commercialization of physical AI and ultimately implement a robot-centric autonomous operating system."
At the event, LG CNS also demonstrated for the first time in Korea heterogeneous robots that autonomously work through training without human remote control. Four types of robots—bipedal, quadrupedal, wheeled, and autonomous mobile robots (AMRs)—that completed training with PhysicalWorks Forge performed tasks in organic collaboration in a logistics setting based on PhysicalWorks Baton.