Development of a robotic welding system for multi-layer multi-pass CJP groove welds between a continuity plate and a steel box column

The steel construction industry faces a critical shortage of skilled welders, driving a transition toward automated manufacturing. Although robotic welding is widely adopted in manufacturing, its application in civil construction is severely hindered by massive assembly tolerances and stringent seismic performance requirements for structural joints, such as complete joint penetration groove welds of continuity plates in steel box columns. To address this challenge, this study presents an adaptive robotic gas metal arc welding (GMAW) system integrated with a data-driven hierarchical planning framework. A comprehensive database consisting of 4,434 weld beads from 270 T-joint specimens was first established by digitizing the heuristic operations of professional welders. Utilizing this database to retrieve suitable parameter combinations, a hierarchical planning framework regulates layer geometry and executes a pass-by-pass volume re-allocation algorithm, thereby adaptively mitigating localized geometric variations. The system's performance was validated through welding tests on nine specimens with plate thicknesses of 25 to 40 mm and highly variable root openings. The welding quality for nine specimens demonstrated 100% compliance with non-destructive ultrasonic testing and the demanding seismic criteria of the AWS D1.8 (2021) structural welding code. This research provides a hierarchical layer-pass planning solution that successfully overcomes tolerance issues in actual manufacturing processes, solves the welder shortage problem, and meets the seismic requirements for continuity plate welding.