A novel self-calibration method with POE-based model and distance error measurement for serial manipulators

Abstract

This paper proposes a novel self-calibration method to calibrate the accurate kinematic parameters of serial manipulators. First, a modified POE-based calibration model was established together with the modeling, identification and correction steps. It eliminates the joint offset errors and the dependent constrained parameters of joint screw errors in traditional POE-based models, which has completeness, continuity and minimality properties simultaneously and avoids normalization and orthogonalization operations. It also eliminates transformation errors of position detection information through adopting distance accuracy definition. Then, focusing on measurement step, the flowchart in a virtual closed-loop form is constructed with an infrared laser beam device and a simple observation surface. Meanwhile, the selection procedure for measurement configurations is figured out to determine the optimal number and the corresponding calibration set. Finally, experimental studies were conducted on a self-developed serial manipulator. The results of self-calibration experiments show that the proposed method is feasible and effective, and the distance accuracy could be improved by 73.64 % after performing optimal type. The results of comparison experiments with traditional external or self-calibration methods indicate that the proposed method could further improve convergence property and achievable accuracy of calibration effects, which also outperforms in terms of simplicity and accuracy.