ABSTRACT
Vision sensing is a key technology to realize on-line detection of welding groove sizes and welding torch relative position and posture parameters during the arc welding process of intelligent production. For the specially designed vision sensor based on combined laser structured lights, an integrated calibration method for its internal parameters is proposed firstly, which improves the efficiency, accuracy and comprehensiveness of internal parameter calibration for a line structured light vision sensor and provides a good foundation for industrial application of the vision sensor. Then, the high precision integrated detection algorithms are derived for the V-groove size parameters and the spatial position and posture (SPP) parameters of the welding torch relative to the welding groove based on a single modulated laser lines image. The algorithms make full use of the data in a single modulated laser lines image, adopting data segmentation and plane fitting to realize the 3D reconstruction of V-groove surfaces and its adjacent workpiece surfaces of planar workpiece, so solving the parameters with high precision. In the verification tests, the relative detection error of V-groove size parameters of planar workpiece is less than 1%, and the relative detection error of SPP parameters of welding torch relative to the welding groove is less than 5%, which separately shows the effectiveness and accuracy of the calibration method and the detection algorithms. This research work provides a good technical support for the practical application of the specially designed vision sensor in the intelligent welding production.