Accurate robot calibration by SAI method through visual measurement

Abstract

The SAI method is a simple and flexible calibration method for robot calibration using twist parameters. The theodolite is usually used in the calibration process, which is expensive and inconvenient to be utilized. This paper aims to investigate how to take advantage of visual measuring system for robot calibration effectively. The main idea of the SAI method is to measure the coordinates of robot end points before and after each robot joint rotations in the world coordination. For the visual measurement, only seven images of robot end points before and after each robot joint rotations are needed, and the coordinates could be obtained by using the image points and the internal-and-external parameters of the cameras. The simulation experiments show that the precision of vision measurement is about 0.13 times that of the robot positioning, and the robot positioning error introduced by the end camera decreases as the accuracy of the visual measurement system increases. Accordingly, two 5DMark II cameras, whose measurement precision of which is of 0.16mm, are used as the measuring instruments to calibrate the robot in real experiment. The experimental results show that the positioning error of the robot calibrated by cameras is smaller than 2mm, which is largely consistent with the simulation result. Therefore, using vision to realize the SAI calibration method on robot not only can simplify the process, reduce the cost, but also can ensure the high precision.