An online method for serial robot self-calibration with CMAC and UKF

• The aim of this paper is to propose an online self-calibration method which is used to estimate the kinematic parameters errors of the serial robot manipulators.
• A position marker and IMU are solidly fixed at the robot end-effector (EE) to determine the position and the orientation in real time.
• FQA and CMAC are used to eliminate the influence of the noises and the measurement errors from the sensors.
• The errors between the actual and the nominal kinematic parameters of the robot manipulator could be identified by the Unscented Kalman Filter (UKF).
• This method only takes several uncomplicated steps but performs with high autonomy and accuracy.

The aim of this paper is to propose an online self-calibration method which is used to estimate the kinematic parameters errors of the serial robot manipulators. In this method, a position marker and an inertial measurement unit (IMU) are solidly fixed at the robot end-effector (EE). The position is determined with a position sensor by tracking the marker and the orientation is measured by the IMU in real time. The Factored Quaternion Algorithm (FQA) is used to represent the orientation in a quaternion. In order to eliminate the influence of the noises and the measurement errors from the sensors, the Cerebellar Model Articulation Controller (CMAC) algorithm is adopted to estimate poses of the robot EE. With the estimated poses, the errors between the actual and the nominal kinematic parameters of the robot manipulator could be identified by the Unscented Kalman Filter (UKF). This method only takes several uncomplicated steps but performs with high autonomy and accuracy. Several experiments are carried out with a GOOGOL GRB3016 robot to verify this method and the results indicate that it is indeed of high convenience, precision and efficiency.