Physical AI company MakinaRocks said on Monday it has completed development of a physical AI-based anomaly detection model for welding robots, a core process in shipbuilding, with HD Korea Shipbuilding & Offshore.

The project is the result of a memorandum of understanding signed last year on using AI to improve robot process efficiency, with a focus on rapidly applying artificial intelligence at shipyard sites.

The company said the task was pursued using an anomaly detection approach based on normal operating data, reflecting the industrial-field characteristic of limited fault-history data. MakinaRocks analysed test-building data to identify key causes and patterns behind major abnormal signs in welding robots. It then upgraded the model based on actual data from 12 welding robots used at shipyard sites.

MakinaRocks CEO Seong-ho Yoon (윤성호) said, "Physical AI, which enables AI to operate by overcoming data limitations in various physical assets at industrial sites, including robots, is a mega technology trend that global manufacturing companies cannot avoid." He added, "Based on this collaboration with HD Korea Shipbuilding & Offshore, we will present a milestone in physical AI technology that can be applied in various ways at shipbuilding industry sites."