[Biz Leader] Mikelorobotics: "Robots That Make Human-Like Decisions... Creating Physical AI for Manufacturing"

Turning Shop Floor Experience into Data
Targeting Challenging Processes Like Surface Treatment Where Automation Has Been Difficult

Park Jangjun, CEO of Micello Robotics, and CTO Choi Junhwa along with other founding members of Micello Robotics are seen in a meeting. Provided by Micello Robotics

"The core of physical artificial intelligence (AI) is not whether a robot looks human, but whether it can understand data generated in the real world and translate that experience into action."

Jangjun Park, CEO of Mikelorobotics, defined physical AI, which has recently become a key topic in the manufacturing industry, in these terms. Mikelorobotics integrates AI with vision and sensor technology in existing industrial robots to automate unstructured manufacturing processes that previously required human judgment. The company focuses on implementing the ‘work intelligence’ required on manufacturing sites as software.

Mikelorobotics was founded in 2022 by robotics and mechanical engineering experts who are PhD holders from Seoul National University and have studied abroad. The founding team visited major industrial complexes and manufacturing sites across the country and discovered that many processes remain difficult to automate. The variation in product shapes and working environments made it challenging for conventional, standardized robot automation methods to be applied.

CEO Park explained, "Traditional industrial robots rely on rule-based automation that continually repeats pre-determined coordinates and programs. If the position or shape of a product changes, the system has to be re-taught. Mikelorobotics overcomes this limitation by using cameras and sensors to recognize products and their surrounding environment, and then adjusts the robot’s paths and actions based on work data."

The first target market chosen by Mikelorobotics is surface treatment, including painting, sanding, deburring, and polishing. Surface treatment is a representative field where automation has been difficult — not only do the position and shape of products vary, but operators must constantly adjust work speed, angles, and force according to each situation.

Mikelorobotics’ competitive edge lies in the data it accumulates in the field and its utilization through the proprietary software platform ‘Michelo Studio’. Rather than simply storing robotic movements, the platform connects and digitizes a product’s 3D shape and position, operator motion, robot posture and speed, process conditions, equipment status, and final quality results. This allows the company to learn ‘what actions were taken in various situations and how those actions affected quality outcomes’.

The core functions of Michelo Studio combine into a single workflow: ‘Motion’ captures experienced operators’ movements as data; ‘Path Generation’ creates work paths based on 3D design drawings or scan data; ‘VisCo’ corrects for deviations in product location and shape through 3D vision; and ‘Inspection’ provides surface inspection to verify results. Chunhwa Choi, Chief Technology Officer of Mikelorobotics, explained that this platform is being designed so not only robot engineers, but also site workers, can use it with ease.

Mikelorobotics is rapidly expanding into various manufacturing sectors, including automobiles, home appliances, and furniture, through collaborative projects. Its major clients include Hyundai Mobis, HD Hyundai Samho, HD Construction Equipment, Sidiz, Susan Industry, Jigeumgang, and Anyoungsa—companies that typically run multi-variety production environments. The company is also broadening its business into general manufacturing, power plants, shipbuilding, and defense, targeting about twice the annual growth each year, beginning with 11 lead projects in 2025.

The company is also scaling up its business. Following sales of 1.26 billion won in 2025, Mikelorobotics is targeting sales of over 3 billion won this year, with plans to reach more than 5 billion won in 2027 and over 12 billion won in 2028.

The critical focus is moving away from simply repeating project-based system integration (SI) business. CEO Park stated, "In the conventional robot SI business model, as sales increase, more engineers are needed. We are now commercializing Michelo Studio so that the path generation methods, correction logic, data structures, and process recipes developed for the first site can be reused at subsequent sites."

To this end, the company plans to expand its standard packages and partner channels starting this year, then transition to a software- and data-driven business. Park cited major goals, including shifting to package sales and partner expansion in 2026-2027, and achieving 20% of revenue from software-as-a-service (SaaS) by 2028.

Recently, business expansion is expected to accelerate with the acquisition of Mikelorobotics by fintech and B2B IT infrastructure firm Finger. Mikelorobotics will collect and process work data from manufacturing sites, while Finger will provide data management and platform-building expertise to manage higher-level systems. The goal is to ultimately expand into an integrated platform that connects traditional manufacturing systems such as enterprise resource planning (ERP) and manufacturing execution systems (MES) with actual work data from the field.

Park stressed, "If Mikelorobotics standardizes the robot’s language, then, through integration with Finger, that language can be used by the entire company as part of its IT infrastructure." Mikelorobotics handles standardization of work paths, postures, speed, and process conditions for robots, while Finger provides corporate IT expertise covering API linkage, data management, and authority and history tracking.

The technology developed in surface treatment will later be extended to fields such as finishing and assembly. Assembly processes like bolting and connecting, which require a lot of human intervention, are seen as prime candidates for physical AI, as they require adaptation to variety in product location and shape.

In the global market, the company will initially focus on Japan and the United States. In Japan, Mikelorobotics aims to secure high-quality reference cases to meet stringent manufacturing standards, while in the U.S., it plans to enter alongside domestic clients’ production hubs in North America. Rather than creating large SI organizations abroad, the plan is to equip local partners with the Michelo Studio platform, offering packages, training, and eventually software development kits (SDKs).

Mikelorobotics’ founders declared, "By 2030, Mikelorobotics intends to transform into a global manufacturing physical AI platform company—turning human expertise on the factory floor into data, which can then be used as work intelligence by any robot."

This content was produced with the assistance of AI translation services.