Physical AI serves as a crucial turning point for the revitalization of South Korea's manufacturing sector. Its significance lies in its ability to operate factories more efficiently and transform the accumulated, long-term experience of skilled workers into future assets. Park Jong-hoon, CEO of Neuromeka, believes that this technology can resolve industry-wide challenges such as labor shortages, the burden of hazardous tasks, and the automation barriers associated with high-mix, low-volume production.

On the 22nd, EToday will host "2027 TECH QUEST - AI LIVE 2027" at the KBIZ Hall of the Korea Federation of SMEs in Yeouido, Seoul, to commemorate its 20th anniversary. CEO Park Jong-hoon will participate as a speaker at the event, sharing the direction of AI technology and manufacturing innovation under the theme "Innovation in Robot Automation through Physical AI."

Founded by CEO Park in 2013 with the goal of bringing robot automation to small and medium-sized enterprises (SMEs) and small business owners, Neuromeka is recognized as a full-stack robotics company that internalizes the entire value chain—from core robot components to Physical AI and on-site automation. The company possesses robust robot automation capabilities, ranging from process analysis and robot EOAT (End-of-Arm Tooling) design to integrated vision control and facility safety interlocking. Currently, it supplies automation solutions to diverse sectors such as steelmaking (POSCO), shipbuilding (HD Hyundai Samho), precision machining (DN Solutions), and F&B (Kyochon Chicken). CEO Park emphasized both the safety and the ease of operation of collaborative robots working side-by-side with humans.

Highlighting that South Korea operates 1,220 robots per 10,000 manufacturing workers, CEO Park stressed, "South Korea is the country that utilizes robots the most in the world." This scale is nine times higher than the global average of 132 robots. In 2021, the country surpassed 1,000 robots per 10,000 workers for the first time globally, and by 2024, the figure increased by over 20% compared to the previous year.

In a manufacturing powerhouse heavily reliant on robots, a significant portion of costs is funneled into "teaching the robot" rather than the hardware itself. CEO Park elaborated, "Looking at the cost composition of a single robot automation project, robot hardware accounts for 25 to 40%, while programming and facility integration take up 50 to 70%." This explains why factories producing large quantities of identical items were the first to automate. He added, "While simple mass production can achieve return on investment (ROI) within a year, complex or high-mix, low-volume production plants require an ROI period of two to seven years."

CEO Park diagnosed that these issues can be resolved using Physical AI. He asserted that robots must acquire the ability to perform diverse tasks by learning extensive data through Physical AI, while simultaneously executing them accurately and precisely based on foundational robotic capabilities. "No matter how good a robot's vision is, without fingertip sensation, a needle cannot thread a thread," Park pointed out. "Robots do not fail because they fail to find an object. After finding it, they fail in the final 1 centimeter. Two-thirds of failures occur in that 'last 1 centimeter.'"

Neuromeka's Physical AI completes tasks through step-by-step verification, status monitoring, and recovery in abnormal situations. Before actual execution, it pre-verifies path collision reachability, and upon detecting a failure, it automatically selects stopping, retrying, or re-planning.

"We plan to expand industrial applications into process-specific packages for manufacturing, laboratories, and medical fields by scaling up our Physical AI platform and skills—such as transitioning from single-arm collaborative robots to dual-arm robots—in a short period," CEO Park stated. Neuromeka is currently conducting Proof of Concept (PoC) tests in automotive electronic and exterior component assembly, as well as logistics transportation processes. The company plans to complete demonstrations this year and verify task intelligence capable of repeated application across diverse production environments.

EToday will diagnose the current state of South Korea's AI industry—covering AI semiconductors, industrial AX, Physical AI and robotics, and smart factories—while exploring the direction of next-generation technologies and industries leading 2027. Experts representing government, academia, and industry will gather in one place to present competitiveness strategies and new growth blueprints for South Korea's AI ecosystem.

[ⓒ Global Economic Times. 무단전재-재배포 금지]