Online robot auto-calibration using IMU with CMAC and EKF

Abstract

One of the possible accurate, efficient, low-cost robot auto-calibration methods is to adopt an Inertial Measurement Unit (IMU) which is rigidly attached to the robot end-effector (EE). The end-effector orientation is measured by calibration of the IMU at every robot measurement configuration. Based on this idea, this paper proposes an online robot auto-calibration method with some developments. In order to eliminate the noise and measurement error of the IMU, a Factored Quaternion Algorithm (FQA) and a Cerebellar Model Articulation Controller (CMAC) algorithm are integrated in use to estimate the orientation of the robot EE. With the estimated orientation, the kinematic parameter errors could be obtained by the Extended Kalman Filter (EKF). Compared to the existing robot calibration methods, this method does not require complex procedures, for example the image capture and process, which makes it more intelligent and efficient. With this method in robot production and maintenance, the reliability and accuracy of the manipulator orientation will increase. To verify the proposed method, several experiments are carried out on a GOOGOL GRB3016 robot and the results indicate that this method is of higher precision, efficiency and convenience than the vision-based methods.