Robust Calibration Approach for Robot Base Coordinate System Based on RANSAC-SVD

Base coordinate system calibration is a critical issue when performing certain tasks in a uniform coordinate system, especially in industrial manipulation. Conventional calibration approaches are susceptible to rotation matrix normalization and measurement noise, which may influence the calibration accuracy and robustness greatly. To address this problem, a robust calibration approach based on random sample consensus (RANSAC)-singular value decomposition (SVD) is proposed in this article. Specifically, a closed chain configuration is first adopted to construct the pose relationship between the manipulator and external measuring equipment. The intermediate orthonormal rotation matrix is obtained using the SVD-based approach with four noncoplanar points. Then, the candidate calibration solution is finally estimated based on RANSAC by alleviating the effects of measurement noise and using the robust iterative scheme. The proposed calibration approach possesses the advantage of not only reducing the impact of measurement noise on the calibration accuracy greatly, but also avoiding the nonorthogonalization problem when finding the solution of the rotation matrix. Then, the numerical simulations are conducted on the calibration accuracy and robustness in the presence of different levels of noise. A research prototype of the calibration system consisting of the industrial manipulator and the binocular camera is developed. Experiments are carried out to validate that the RANSAC-SVD-based calibration approach is superior to other calibration approaches in terms of calibration accuracy and robustness. The calibration approach can provide a reference for other pose calibration problems.