Wonpredict Joins National Physical AI Standardization Project

Recap

Source: 로봇신문, report of Oct. 6, 2026

According to the report, Wonpredict, a manufacturing AI specialist led by CEO Yoon Byung-dong, has joined a large-scale national research and development project. The initiative, titled 'Creating an R&D Ecosystem for Collaborative Intelligence Physical AI-Based SW Platforms,' is driven by the Ministry of Science and ICT and the National Information Society Agency (NIPA). The total budget for the broader program is 736.8 billion KRW, with 515 billion KRW in government funding allocated through 2030. A physical AI demonstration hub is also planned for Wanju, Jeollabuk-do.

Wonpredict is participating in a specific sub-project led by Chonbuk National University's Industry-Academia Cooperation Foundation, focused on standardizing software-defined factory operation and control systems (SDF-OCS). This particular task has a budget of approximately 72 billion KRW and runs from August 2026 to December 2030, involving 13 institutions including universities, standardization bodies, and software firms. The core goal is to create a flexible manufacturing software system where equipment, robots, and AI models can be swapped out without requiring a complete overhaul of the factory's control architecture.

The company’s specific role involves developing a 'Common Manufacturing Framework' that standardizes how AI operates on the shop floor. This includes creating six general-purpose AI frameworks: intelligent anomaly diagnosis, predictive quality analysis, autonomous predictive maintenance, adaptive production planning, autonomous process control, and self-correcting quality. Wonpredict aims to use foundation models and transfer learning techniques to create specialized models for individual factories using fewer than 1,000 data points. These modules will be cross-validated across different industries, such as casting, food processing, and precision parts manufacturing.

Context

This project addresses a persistent pain point in industrial automation: the lack of interoperability. Traditionally, changing a robot model or a piece of equipment often required rewriting application software or modifying the top-level control system. The SDF-OCS concept seeks to solve this by establishing common protocols and standard interfaces, allowing manufacturers to treat AI capabilities as interchangeable modules rather than bespoke, one-off solutions. This aligns with the broader global trend toward 'Physical AI,' where digital intelligence is tightly integrated with physical hardware to enable autonomous decision-making in real-world environments.

Wonpredict has previously applied AI for predictive maintenance and quality prediction in various energy and manufacturing sites. This national project represents a shift from site-specific solutions to a standardized, reusable platform. The planned public catalog, FacHub, is designed to lower the barrier to entry for small and medium-sized enterprises (SMEs) that may lack the resources to develop their own AI infrastructure.

Robot's take

The significance of this initiative lies in its potential to decouple AI capabilities from specific hardware, a critical step toward scalable industrial intelligence. If the SDF-OCS standard gains traction, it could significantly reduce the cost and time required for factories to adopt advanced AI, moving beyond the current model where each implementation is a unique, expensive project. However, the success of such a standard depends heavily on industry-wide adoption and the ability to truly abstract complex manufacturing processes into modular components. It remains to be seen whether the 'less than 1,000 data points' claim for model adaptation will hold up across the wide variance of industrial environments, but the cross-validation plan across different sectors is a promising approach to testing this robustness.