The problem of recognizing and locating rigid objects in 3-D space is important for applications of welding automation. The issue relates to finding transformation matrix between the planned location and actual one to compensate the fixturing errors before welding starts.
In this study, a vision sensor based on the optical triangulation was used to collect the range data of die surfaces for automatic remodeling. In measuring with the laser vision sensor, erroneous range measurements can be caused by a specular die surface. Thus, the specular reflection was analysed by geometrical equations to calculate the range of specular reflections. The measuring experiment was performed within the non-specular reflection range.
Line and plane vector equations were constructed by the measured range data, and an analytic algorithm was proposed for recognizing the die location with these vector equations. This algorithm could make the transformation matrix without any specific corresponding points. To ascertain this algorithm, folded SUS plate was measured by the laser vision sensor attached to a 3-axis cartesian manipulator and the transformation matrix was calculated.
The recognizing errors of the SUS plate location were all within 2mm. These errors are considered to be allowable for die remodeling. In the future, other recognizing algorithms should be developed for the curved surfaces.
