Kick-off Meeting for Physical AI R&D Project

"Launching Physical AI R&D and Demonstration Initiatives"

The government will invest approximately 1.4 trillion won to promote research and development (R&D) as well as demonstration projects for physical artificial intelligence (AI) in the regions of North Jeolla Province and South Gyeongsang Province. The plan is to establish standard technologies for physical AI, such as software and models, through demonstrations in regions with a strong manufacturing base.


The Ministry of Science and ICT held a kick-off report meeting for the physical AI R&D project on the morning of September 16 at the Novotel Ambassador Hotel in Jung-gu, Seoul, focusing on the manufacturing sites of North Jeolla Province and South Gyeongsang Province.


Government to Invest 1.4 Trillion Won in Securing Physical AI Technology for Manufacturing in North Jeolla and South Gyeongsang Provinces View original image

The Ministry plans to establish an end-to-end physical AI R&D framework by linking manufacturing data, AI models, digital twins, robot and equipment control, and field demonstration, with a focus on North Jeolla and South Gyeongsang Provinces. The goal is to develop and demonstrate technologies that can operate in real-world manufacturing environments, leveraging the distinct technical strengths and manufacturing sites in these regions.


In North Jeolla Province, the project titled "Research and development of a collaborative intelligence physical AI-based software platform and ecosystem creation" will focus on developing and demonstrating core physical AI technologies specialized for manufacturing and logistics. A total of 736.8 billion won will be invested into this project over the next five years, until 2030.


Through this initiative, a physical AI software platform enabling collaboration among different robots and facilities will be developed and verified in real manufacturing environments. Beyond collaboration among heterogeneous robots and facilities, the project will also take on the challenge of "factory orchestration," integrating the entire factory into one system. This involves connecting automated factory layout design, logistics robot route planning, 3D digital-twin-based virtual verification, vision-language-model (VLM)-based situation awareness, and reinforcement-learning-based integrated control to real testbeds and field demonstrations.


Based on the results of last year's proof of concept (PoC) that showed applicability in real manufacturing settings, the plan is to secure a general-purpose dark factory model using AI for factory design, construction, and operation, further developing it into an export model for K-manufacturing factories equipped with standardized and interoperable features.


In South Gyeongsang Province, which has a strong manufacturing base in industries such as defense and shipbuilding, the project "Development of human-AI collaborative large action models (LAM) and global demonstration" aims to develop physical AI technologies specialized for precision control that will enable collaboration between humans and AI in manufacturing sites—including the actual execution of tasks. A total of 676.3 billion won will be invested in this project over the next five years, until 2030, focusing on high-quality manufacturing data acquisition, physical AI technology development, and demonstration of production processes.


The project will develop LAM models capable of precision control and real-world actions by integrating data on human and robot behaviors at manufacturing sites with physical laws and spatial information. Furthermore, it will create a cyclical structure by linking digital twins, data pipelines, simulation, and infrastructure based on neural processing units (NPUs), facilitating learning, on-site deployment, and iterative re-training of models.


By next year, the target is to secure representative manufacturing-process data, key data pipelines, bi-directional digital twins, minimum viable prototypes (MVPs) of the platform, and major validation systems. By 2030, the project aims to establish field digital twins and virtual manufacturing factories (FAB) for more than 13 manufacturing processes, with the goal of achieving a closed-loop automation rate of over 90% and reflecting retraining within 24 hours through integrated demonstration efforts.


More than 3,500 researchers from around 150 universities, research institutes, and companies will participate, with 107 small and medium-sized enterprises accounting for about 68% of the total.


This kick-off meeting is significant in that it has activated a virtuous cycle where research technologies are applied to real manufacturing sites and the data collected onsite is used to further advance AI models and operating systems. The event was attended by attendees including Lee Dokyu, policy chief at the Ministry of Science and ICT; Unification Minister and National Assembly member Chung Dongyoung; Assemblyman Choi Hyungdoo; Go Jeongsam, Vice Governor for Economic Affairs of North Jeolla Province; Choi Manrim, Vice Governor for Economic Affairs of South Gyeongsang Province, and about 200 others. The Ministry shared the project direction and a five-year roadmap for 2026–2030.



Lee Dokyu, policy chief at the Ministry of Science and ICT, stated, "The government will focus its policy capabilities on the area of physical AI, one of the three major megaprojects, to take the global lead," adding, "We will actively support the technological developments and on-site demonstrations starting in North Jeolla and South Gyeongsang Provinces so that their achievements spread to manufacturing sites nationwide and ultimately serve as a new source of industrial competitiveness through the global export of Korea's 'K-manufacturing factory' model."


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

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing