基於焊道幾何進行埋弧焊焊炬軌跡自動規劃

This study combines stereo vision and geometric analysis technology to construct a strategy for automatic torch trajectory planning of submerged arc welding of a mobile gantry system, which can be applied to the multi-pass welding of large-scale steel column structures. The geometric description of a weld bead can be obtained from the point cloud generated from serial depth images. Welding point coordinates can be determined based on the ordinal point cloud profiles of the weld bead, obtaining the movement trajectory of the welding torch. Given the preliminary test, the feasibility has been verified, in which the RMSE of the welding points is about 0.09 mm, achieving the sub-pixel accuracy of the imaging sensor. Also, with a parallel computing framework, the computational efficiency increases up to 18%, suggesting the proposed strategy can assist the automation of submerged arc welding and reduce labor costs.