Abstract
This paper proposes a novel approach for automated planning of robotic MAG welding processes based on an adaptive gap model. The gap model describes the relation between welding parameters and a changing gap geometry resulting from deviations of relative part locations in a welding assembly. A matching process is performed between computer-aided design model and a measured point cloud of the welding assembly using an Iterative Closest Point algorithm to compute the actual gap geometry. Experimental validation for various gap geometries is demonstrated with an industrial robot equipped with stereo camera and welding gun indicating an increased weld quality.