Deflection model for robotic friction stir welding

Purpose - This paper aims to present a deflection model to improve positional accuracy of industrial robots. Earlier studies have demonstrated the lack of accuracy of heavy-duty robots when exposed to high external forces. One application where the robot is pushed to its limits in terms of forces is friction stir welding (FSW). This process requires the robot to deliver forces of several kilonewtons causing deflections in the robot joints. Especially for robots with serial kinematics, these deflections will result in significant tool deviations, leading to inferior weld quality. Design/methodology/approach - This paper presents a kinematic deflection model, assuming a rigid link and flexible joint serial kinematics robot. As robotic FSW is a process which involves high external loads and a constant welding speed of usually below 50 mm/s, many of the dynamic effects are negligible. The model uses force feedback from a force sensor, embedded on the robot, and predicts the tool deviation, based on the measured external forces. The deviation is fed back to the robot controller and used for online path compensation. Findings - The model is verified by subjecting an FSW tool to an external load and moving it along a path, with and without deviation compensation. The measured tool deviation with compensation was within the allowable tolerance for FSW. Practical implications - The model can be applied to other robots with a force sensor. Originality/value - The presented deflection model is based on force feedback and can predict and compensate tool deviations online.