An Optimal Path-Tracking Algorithm for Unstructured Environments based on Uncalibrated Vision

Abstract

This paper presents aspects related to the development of a technique for three-dimensional path tracking over unstructured surfaces. Path tracking is performed by an industrial robot, using a vision-based technique known as camera-space manipulation. In this work a previously-defined, three-dimensional trajectory is placed, in an optimal fashion, over a curved, unstructured surface. The method was tested with an industrial Fanuc ArcMate 100i robot, together with structured lighting in the form of a matrix array of laser beams. Such source of light provides a way to facilitate the image-analysis process required to characterize the surface where the task is going to be performed. A demanding industrial task such as welding was performed with the methodology presented herein. Experimental results show a complex path welded over a surface of unknown geometry.