Automatic path planning for welding robot based on reconstructed surface model

An automatic path planning method, including the torch pose planning, is proposed to improve the flexibility and automatic planning ability of arc welding robot. The virtual object, which representing the real object to be welded, is reconstructed firstly by non-rational B-spline surface from geometric polygonal mesh model or measured point cloud. Based on the reconstructed non-rational surface model, the tangent vector and the tangent plane of each sampling point on the space welding seam is calculated, as well as the normal plane of each sampling point. Thus the position and pose of torch with respect to each sampling point of space seam can be determined by the calculated tangent vector, normal plane and concrete requirement of welding process. The final path planning of arc welding robot in joint space is obtained by its inverse kinematics. Experimental results demonstrate the accuracy and the feasibility of the proposed method.