A Simultaneous Calibration Method for Dual-Manipulator System Based on Holistic Kinematic Error Model

Abstract

The dual-manipulator system is crucial on space assembly tasks but struggles with the complexity and error accumulation of calibrating individual components separately. In this article, we propose an accurate and efficient method to achieve simultaneous calibration without multiple calibration process for all the components. First, we derive a holistic kinematic error model, including the two manipulator's link parameters, as well as the flange–camera, flange–marker, and base–base transformations. It is formalized as an extension of the single-manipulator kinematic error model, making it easily compatible with identification algorithms. Second, we design a hierarchical identification algorithm based on Levenberg–Marquardt algorithm and covariance matrix adaptation evolution strategy algorithms, which improves the solution divergence caused by the high-dimensional nonlinear characteristics of the holistic kinematic error model. The physical experiment shows that, compared to the nominal parameters, our method improves translation and rotation accuracy by 77.2% and 57.9%, respectively. It reduces the number of calibrations for all components compared to multiple separate calibration schemes and achieves higher accuracy.