A Humanoid Robot that walks on two legs lifted a box placed on a conveyor belt and turned its body to the left. When the humanoid passed the box to a wheeled quadrupedal logistics transport robot, the transport robot moved by Autonomous Driving and handed it to a wheel-type humanoid standing on the opposite side. The wheel-type humanoid, which has wheels instead of two legs, used its long arms to lift the box and stack it on a shelf more than 2 meters high.
This is a look at a future logistics site demonstrated by LG CNS. The three robots operated based on LG CNS's robot transformation (RX) platform Physical Works and performed tasks autonomously without human intervention. An LG CNS official said, "It is a scene where the robot works by making 100% of its own decisions after learning based on the Robot Foundation Model (RFM), the brain of the robot," and noted, "In logistics sites, there are many tasks that require grasping and placing items of various shapes, and robots made by different manufacturers recognize the items, decide for themselves how to pick them up, and collaborate autonomously."
LG CNS said on the 7th that it aims to foster the robot software field as a new growth engine and leap forward as a leading RX company. Although it does not make robot bodies, it plans to provide an artificial intelligence (AI)-based platform that enables robots to perform tasks automatically in manufacturing and logistics sites. The goal is to help robots understand the site and move on their own to increase the productivity of corporations and lower expense.
On this day, LG CNS unveiled its RX platform Physical Works, which will be the mainstay of its robot business expansion. It is the first time a domestic corporations has introduced, under its own brand, an end-to-end (E2E) RX platform that integrates the entire process from robot learning to operation.
Hyun Shin-Gyoon, president of LG CNS, said at the RX Media Day held at LG Science Park in Magok, Seoul, on the 7th, "Robots are evolving beyond simple automation equipment into agents that actually carry out production and operations," and added, "Amid this change, the competitiveness of corporations depends on how quickly they apply robots to the site, operate them stably, and connect them to results through consolidation."
Until now, the robot industry has developed around hardware performance, but the explanation is that performance in industrial sites is not sufficient with robot hardware alone, and it is important to apply software that allows robots to understand actual tasks and perform them stably.
Hyun said, "The core of robot transformation is not the performance of individual robots but establishing a learning, verification, and integrated operation system suited to the site," and added, "As LG CNS has accumulated experience by carrying out hundreds of intelligence and automation projects in logistics and manufacturing sites, applying and operating various robots, we aim to become a key partner that supports the entire RX process for customers going forward."
LG CNS plans to go all out to target the RX market starting in the second half of this year. To this end, it has been developing its own robot operation and learning platform, Physical Works, and this year made a strategic investment in the U.S. Humanoid Robot company Dexmate. Previously, it invested in the U.S. robot brain developer Skilled AI and is also collaborating with Config, an RFM company specializing in dual-arm humanoid control.
LG CNS cited as the greatest strength of Physical Works the ability to centrally control and manage various types of robots made by different manufacturers in one place. In industrial sites, different forms of robots—bipedal, quadrupedal, and wheel type—must be used depending on the type of work, but until now, control methods and operation screens differed by manufacturer, making efficient management harder as the number of robot types increased. Physical Works standardized robot operating status and control information to solve this problem.
Physical Works is divided into Physical Works Forge, which supports data learning and verification, and Physical Works Baton, which operates robots from different manufacturers such as bipedal, quadrupedal, and wheel type. Based on the two platforms, LG CNS said it has reduced the time required for intelligent robots from learning to field deployment from several months to 1–2 months.
It also projected that applying Physical Works Baton to a robot operation environment of around 100 units, such as autonomous mobile robots (AMR) and automated guided vehicles (AGV), would increase productivity by about 15% or more and reduce operating costs by up to 18%. The company said the effect will be greater the more mixed the environment is with robots from various manufacturers, as duplicate movement, congestion, and manual intervention decrease.
LG CNS is currently conducting robot proof-of-concept (PoC) projects with 20 customers in major industrial sectors. Physical Works Baton is being used to centrally manage four types of robots—patrol, barista, luggage-carrying, and cleaning—in the Busan Smart City national pilot city project. There are no robots yet being operated in actual industrial sites, and the company projected it will take about two years from actual application in the field to producing results.
Hyun cited as LG CNS's unique strengths its production IT system capabilities built up over 40 years and its deep understanding of the field. The plan is to apply this to Physical Works and build an RX platform optimized for manufacturing and logistics sites. The company first intends to apply it to domestic manufacturing and logistics sites, advance it, and then expand to other fields. After securing application cases for Physical Works and advancing them, it also plans to build additional industry-specific RFMs.
Hyun said, "Existing industrial robots focused on automation that repeatedly performed predefined tasks, but in the physical AI era, robots with intelligence perceive situations on their own, make judgments, and carry out tasks autonomously," and added, "We will establish robot introduction strategies optimized for customer sites, set a new standard for the commercialization of physical AI, and ultimately implement an autonomous operation system centered on robots."