Vision-assisted robotic finishing of friction stir-welded corner joints

One required process in the fabrication of large components is welding, after which there may be a need for machining to achieve final dimensions and uniform surfaces. Friction stir-welding (FSW) is a typical example after which a series of deburring and grinding operations are carried out. Currently, the majority of these operations are carried out either manually, by human workers, or on machine tools which results in bottlenecks in the process flows. This paper presents a robotic finishing system to automate the finishing of friction stir-welded parts with minimum human involvement. In a sequence, the system can scan and reconstruct the 3D model of the part, localise it in the robot frame and generate a suitable machining path accordingly, to remove the excess material from FSW without violating process constraints. Results of the cutting trials carried out for demonstration have shown that the developed system can consistently machine the corner joints of an industrial scale part to desired surface quality which is around 1.25 μm in, Ra, the arithmetic average of the surface roughness.