The South Korean government is implementing a 'moving target' approach for its Gyeongnam and Jeonbuk Physical AI research and development (R&D) project, part of a larger mega project. This strategy allows executing agencies to propose adjustments to their plans based on detected technological changes.
The Ministry of Science and ICT and the National IT Industry Promotion Agency (NIPA) held a briefing on the '2026 Gyeongnam and Jeonbuk Physical AI R&D Project' on July 10 at Nuri Dream Square in Mapo-gu, Seoul.
A NIPA official stated, "If the direction of technology changes, we cannot simply follow the existing plan. If the proposed moving target operation method is deemed feasible, we will allow changes to the execution plan after selection." The official added, "If adjustments to the budget or participating organizations are necessary due to the moving target, we will present the issue to the project consultation body for review and incorporate feedback into the execution plan if consensus is reached."
The government's decision to adopt this method stems from the fact that physical AI has been around for less than two years. This reflects the rapidly changing research landscape, as technologies that seem promising at the start of the project may lose competitiveness by its conclusion. The NIPA official explained, "The aim is to operate the rigid execution plans of existing national R&D more flexibly."
As a result, participating consortia must propose a system to continuously track technological trends through papers, technical reports, news, and researcher communities. They must also outline how they will adjust models, research methods, and organizational structures in response to confirmed technological changes. The NIPA official emphasized, "The specificity of the moving target response system will impact the project selection evaluation in the future."
The Gyeongnam and Jeonbuk Physical AI R&D projects focus on different areas. Gyeongnam aims to establish an environment where humans and AI collaborate in manufacturing facilities that are already operational. The project involves collecting the know-how of skilled workers and equipment data from existing manufacturing processes and developing physical intelligence behavior models and world models by combining this data with synthetic data reflecting physical laws.
In contrast, the Jeonbuk project focuses on the integrated operation of unmanned factories with minimal human intervention. Rather than developing individual robots like humanoids or grippers, the project will create a software-defined factory (SDF) and operational control system (OCS) that connects multiple robots, equipment, sensors, and logistics systems.
Additionally, the project plans to establish a joint research and development center, a neural processing unit (NPU) technology verification station, and a physical AI verification meta-factory to validate unmanned factory technologies. The long-term goal is to develop a Korean reference model capable of exporting at the factory level.
However, concerns have been raised that the centralized project structure may pose barriers to entry for small and medium-sized enterprises (SMEs). Companies with the technical capability to execute detailed tasks must still form large consortia with a lead organization and other participants. One company representative at the briefing remarked, "While we have the technical expertise for specific tasks, the burden of forming a consortium for the lead project is significant. Are there opportunities for connecting institutions that can take on the lead role?"
In response, NIPA officials stated, "We are currently unaware of which consortia are being prepared and direct support for formation is difficult. We encourage finding collaborative partners through regional technoparks and the Physical AI Alliance."
Meanwhile, major companies such as Naver Cloud, Megazone Cloud, LG CNS, KT, and Siemens, along with AI firms like Persona AI, Real World, Flitto, and Aim Intelligence, attended the event. The project announcement will be open until July 28, with the application period running from July 17 to 3 p.m. on July 28. NIPA plans to conduct evaluation presentations from August 5 to 7 and select executing agencies by the end of August to commence research and development.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.