NC AI said on Wednesday it has won a Hanwha Ocean project titled "Development of a vision-recognition-based welding-only model and a collaborative-robot-based autonomous welding model" and has launched a cooperation framework.

The project combines artificial intelligence vision recognition and precision robot control technology for welding, a core shipbuilding process. The aim is to build an autonomous welding physical AI solution in which robots recognise and judge welding areas on their own and perform optimal welding in real time. The development output will later be applied in practice to Hanwha Ocean’s commercial and special-purpose shipbuilding processes.

Shipyard welding sites are environments in which vision-recognition AI is difficult to operate due to arc light and sparks, welding dust, and lens contamination. NC AI is developing shipbuilding-specific vision recognition technology that precisely extracts weld lines and detects defects in real time even amid noise and contamination, by reflecting Hanwha Ocean site data and engineer feedback in real time.

NC AI plans to deploy its next-generation industry-specialised vision-language model (VLM), "VAETKI Vision", as the core engine for the project. VAETKI Vision is an industry-focused model that integrates visual information and text instructions, and NC AI will expand it into a vision-language-action (VLA) model. Under this approach, when a worker gives instructions by voice or text, the collaborative robot visually analyses the shape of the welding target and the condition of the weld line and derives control commands such as torch angle and speed on its own.

Lee Yeon-su (이연수), chief executive of NC AI, said, "We will develop robust vision recognition technology and autonomous control models that overcome dust and contamination on site to create a physical AI model at the world’s highest level."