On the comparison of model-based and modeless robotic calibration based on the fuzzy interpolation technique

Abstract

Traditional robot calibrations implement model-based and modeless methods. The calibration of position error in model-based method implements four steps: kinematic model definition of the robot, measurement process of the robot's positions, identification of the kinematic model of the robot and compensation of the position errors. This method is both time consuming and complicated because of the identification and compensation processes. Most traditional modeless methods are relatively simple and practical, but the calibration accuracy is relatively low and most of them are only suitable for the robots calibration in 2D domain and in relative small workspaces. This paper provides a novel modeless fuzzy interpolation method to improve the compensation accuracy for the robots calibration in 3D workspace. A comparison between the model-based and modeless fuzzy interpolation calibration method is made. The simulated results show that the modeless fuzzy interpolation calibration method is compatible with the model-based calibration method, and even outperforms a model-based counterpart in terms of accuracy and complexity in small workspaces.