Measurement and Trace Generation of a Friction Stir Welding Robot for Space Weld Joint on Large Thin-Walled Parts

For the large thin-walled metal part, it is difficult to determine the spatial weld curve when it deforms due to manufacturing or clamping, especially the spatial normal vectors of the part. To solve this problem, a method is presented to estimate the spatial normal vector and calculate cutter location points by measuring the discrete points on the weld joint. Firstly, a contact-type probe fixed in the end of FSW (friction stir welding) robot is used to touch the weld joint to measure a series of space points. By fitting these points with a cubic spline curve, a spatial curve which is one probe radius long distance from the real weld curve can be obtained. Secondly, the least square plane of the measured points is solved by the least square method. Then, the planar normal vectors of the projection curve in the least square plane are used to estimate the spatial normal vectors of the real weld joint curve. At last, fictitious cutter location points discretized by the chord height error algorithm on the fitted curve are moved along the direction of their space normal vectors by one probe radius distance to get the real cutter location points. Simulation experiment demonstrates the effectiveness of the proposed method. Using the method, cutter location points for welding and normal vectors can be obtained without measuring the entire surface of the large thin-walled complex rotating parts in FSW.