Circular test, positioning accuracy and repeatability of industrial robots using extrinsic parameters of a camera

This document presents the experimental validation of image processing techniques used to estimate displacements for end-effector robot or worktable for machine-tools. The aim is performance evaluation of robots and machine-tools. Since the most important performance parameters to be evaluated are accuracy and repeatability, we need to assure that accuracy of image processing techniques is at least twice smaller than mean accuracy expected for machines. Two main characteristics should be carefully revised: camera calibration accuracy and pose estimation accuracy. To validate this technique we used a linear stage and a rotary stage, for linear and angular displacement respectively. Finally, the performance evaluation of a spherical parallel mechanism designed for milling applications is presented. The circular test for different angles from the vertical was performed. A grid checkerboard pattern was fixed to end-effector. Circular motions are executed while a high resolution, previously calibrated, camera acquires images of the pattern. The end-effector pose is estimated from the extrinsic parameters of the acquired images.