A New Variable Step-Size Levenberg-Marquardt Algorithm for Industrial Robot Calibration

Abstract

Industrial robots are a critical equipment to achieve the automatic production, which have been widely employed in industrial production activities, like handling and welding. However, due to some inevitable impact factors such as machining tolerance and assembly tolerance, a robot suffers from low absolute positioning accuracy, which cannot satisfy the requirements of high-precision manufacture. To address this hot issue, a new robot calibration method incorporating an unscented Kalman filter with a variable step-size Levenberg-Marquardt algorithm is proposed. The main ideas of this paper are as follow: a) developing a novel variable step-size Levenberg-Marquardt algorithm to addresses the issue of local optimum in a Levenberg-Marquardt algorithm; b) utilizing an unscented Kalman filter to suppress the measurement noises; and c) proposing a novel calibration method based on an unscented Kalman filter with a variable step-size Levenberg-Marquardt algorithm. Moreover, the empirical studies on an ABB IRB 120 industrial robot demonstrate that the proposed method obtains much compared with state-of-the-art methods, the proposed method further outperforms each of them in terms of calibration accuracy for robot calibration. Therefore, this study is an important milestone in the field of robot calibration.