Subordinate regulations under the AI Basic Act will be enacted with a focus on minimizing regulation. By setting a guidance period of at least one year for administrative fines, it aims to create an effect of delaying regulation. [Photo: Shutterstock

The government will begin physical AI research and development focused on manufacturing sites in Jeonbuk and South Gyeongsang provinces. In Jeonbuk, it will develop “factory orchestration” technology that integrates different robots and equipment into a single system. In South Gyeongsang, it will develop a large action model (LAM) capable of real actions and precision control in manufacturing sites.

The Ministry of Science and ICT on Sept. 16 held a project kickoff briefing for a physical AI research and development programme at the Novotel Ambassador in Seoul, focused on manufacturing sites in Jeonbuk and South Gyeongsang.

About 200 people attended, including Lee Do-gyu (이도규), director general for ICT Policy at the ministry; lawmakers Jeong Dong-young (정동영) and Choi Hyung-du (최형두); and Go Jeong-sam (고정삼), Jeonbuk’s vice governor for economy, and Choi Man-rim (최만림), South Gyeongsang’s vice governor for economy, along with officials from research institutes, universities, companies and related organisations.

The ministry will build an end-to-end physical AI R&D system by linking manufacturing data and AI models, digital twins, robot and equipment control, and on-site demonstrations.

In Jeonbuk, it will pursue a project to develop a collaborative-intelligence physical AI software platform and build an ecosystem. It will invest a total of 736.8 billion won over five years from this year to 2030 to develop and demonstrate core physical AI technologies tailored to manufacturing and logistics sites.

The core is a physical AI software platform that enables different robots and equipment to work together. The ministry will use it to implement “factory orchestration” that integrates an entire factory into a single system.

It will automatically design factory layouts and logistics robot routes and verify them in a virtual environment using a 3D digital twin. It will also apply vision-language model-based situational awareness and reinforcement learning-based integrated control technologies to real testbeds and manufacturing sites. Based on proof-of-concept results from last year, it will also secure a general-purpose “dark factory” model in which AI designs, builds and operates factories. It plans to develop this into an export model for “K-manufacturing factories” with standards and interoperability, together with anchor companies by industry.

In South Gyeongsang, it will pursue a project to develop a human-AI collaborative LAM and conduct global demonstrations. It will invest a total of 676.3 billion won over five years from this year to 2030, using manufacturing industry foundations such as defence, shipbuilding and machinery and 210 industrial complexes.

The core technology is the large action model. It is a model that can perform real actions and precision control by learning from data that combines worker and robot behaviours in manufacturing sites, physical laws and spatial information. It will also link digital twins, data pipelines, simulations and neural processing unit-based infrastructure. Through this, it will build a circular system that runs from AI training to deployment at manufacturing sites and retraining using work results.

By next year, the ministry will secure representative manufacturing process data, key data pipelines, a two-way digital twin, a minimum viable product platform and major verification systems. In 2030, it will pursue integrated demonstrations targeting at least 13 manufacturing processes, with goals including building on-site digital twins and a virtual manufacturing factory (FAB), achieving at least 90 percent closed-loop automation, and reflecting retraining results within 24 hours.

About 3,500 researchers from about 150 domestic universities, research institutes and companies will take part in the programme. Small and medium-sized and mid-sized companies account for 107 firms, or about 68 percent of participating organisations. The ministry will secure physical AI technology by using the different manufacturing foundations and technical strengths of Jeonbuk and South Gyeongsang and spread results to manufacturing sites nationwide.

Lee Do-gyu, director general for ICT Policy at the ministry, said, “We will actively support the spread of results from Jeonbuk and South Gyeongsang to manufacturing sites nationwide and ultimately enable South Korea’s ‘K-manufacturing factories’ to be exported to global markets.”

Keyword

#Ministry of Science and ICT #Jeonbuk #South Gyeongsang #physical AI #large action model
Copyright © DigitalToday. All rights reserved. Unauthorized reproduction and redistribution are prohibited.