[Edaily Reporter Shin Yeong-bin ] The government has finalized the specific direction for research and development (R&D) on core physical artificial intelligence (AI) technologies to be pursued next year.

According to the information technology (IT) industry on the 1st, the Institute for Information & Communications Technology Planning & Evaluation (IITP) is reportedly preparing five to six R&D projects to develop core Physical AI technologies next year. It is currently in the process of drafting detailed requests for proposals (RFPs).

Next year’s Physical AI R&D projects are being reviewed with the following as key pillars: △ a general-purpose Physical AI Foundation Model (PFM), △ data pipelines, △ safety, and △ simulation technologies.

Next year’s projects will be carried out in conjunction with the Physical AI Leading Technology Development Project launched this year. The government has been developing World Model core technologies and Physical AI models since this year and views these efforts as preliminary projects leading up to large-scale core technology development initiatives, such as the development of a general-purpose PFM next year.

Data, another bottleneck for Physical AI, will also be addressed. Research will focus on refinement technologies to process collected data into a format suitable for model training and to improve training efficiency. Simulation technologies for generating training data in virtual environments and transferring it to real-world settings are also being considered as key tasks.

Safety is also being examined as a separate priority. This is because, given the nature of Physical AI—which interacts with people and equipment in real-world spaces—errors in model judgment could lead to physical accidents. The government previously outlined policies in its Physical AI strategy for ensuring safety in field applications and developing related technologies.

Government investment in Physical AI will expand significantly starting next year. The total government budget for Physical AI will increase from approximately 900 billion won this year to 3.1 trillion won next year. Field demonstrations in eight key sectors—including manufacturing, construction, agriculture, and defense—along with the adoption of domestically produced AI robots, the development of core technologies, and the establishment of data infrastructure will be pursued simultaneously.

The IITP plans to announce the project call in January of next year, following a public consultation on the RFP in November. It is currently reviewing a schedule to select implementing agencies in March and commence research and development in April.

An IITP official stated, “Currently, most Physical AI foundation models are closer to specialized models than general-purpose ones,” adding, “Our goal is to develop models that can be used universally, regardless of form factor or scenario, rather than for specific tasks such as those in particular factories or logistics.”