A feedforward deflection compensation scheme coupled with an offline path planning for robotic friction stir welding

Abstract

Robotic friction stir welding is an innovative process which allows the welding of aluminum workpieces with robots. With their six-axis, serial industrial robots allow the welding of three-dimensional complex geometries but, under high load induced by the process, they undergo deflection due to their limited stiffness. Position and orientation deviations of the end effector are observed during welding experiments leading to weld defects. In this paper, a feedforward compensation technique based on a deflection model is implemented to compensate the deviations. This technique is coupled with an offline path planning methodology based on Bézier curves. The methodology is validated by performing an arc of circular helix welding path. High position and orientation accuracy are achieved and defect-free weld is performed.