Intelligent Welding System for Aluminum Coils by Adaptive Path Planning with Real-world Dataset

Intelligent welding systems (IWS) are gradually replacing manual welding in industrial production due to their ability to respond in real time to changes in the welding process interface and adjust welding parameters accordingly. However, IWS has primarily been used for welding flat materials, and effective solutions for welding coils are still lacking. This is mainly because welding coils involves complex three-dimensional positioning and planning, which places higher demands on the feedback capabilities of IWS. To address this gap, we propose an smart Al coil welding system(SAWCS). SAWCS combines 2D imaging, 3D point cloud data, and deep learning-based segmentation to identify weldable areas with high accuracy. A dedicated dataset, HRALCoil, was developed to train models for robust feature recognition and adaptive path planning. Through real-world tests, SAWCS demonstrated improved precision and reliability in weld path planning, meeting industrial demands for efficient and stable welds. This approach sets a foundation for expanding intelligent welding applications in manufacturing. Future work will focus on enhancing system adaptability under varying production conditions.