Kinematic calibration of over-constrained robot with geometric error and internal deformation. (July 2023)

• Record Type: Journal Article
• Authors: Ye, Hao; Wu, Jun; Huang, Tian
• Is Part Of: Mechanism and machine theory. Volume 185(2023)
• Publication Date: 2023-07
• Subjects: Internal deformation -- Kinematic calibration -- Over-constrained robot -- Geometric error -- Least action principle; Machine theory -- Periodicals; Machinery -- Periodicals; Machines -- Périodiques; Génie mécanique -- Périodiques; Machine theory; Machinery; Periodicals; 621.81
• DOI: 10.1016/j.mechmachtheory.2023.105345
• Languages: English
• ISSNs: 0094-114X

Abstract

Highlights: General nonlinear calibration framework of robot based on POE formula. Geometric error model of over-constrained robot considering internal deformation. Explicit interaction between internal deformation and geometric errors. Pose errors decrease by about 95% after calibration in a 5-DOF hybrid robot. Abstract: In the over-constrained robot with geometric errors, internal forces would occur and lead to deformation due to redundant constraints and common constraints. In current kinematic calibration approaches, since the deformation, geometric errors and pose errors cannot be described in a unified framework, the error propagation fails to be solved intuitively and generically. In this article, a general calibration framework of over-constrained robots is proposed to unify geometric errors and internal deformation through the least action principle. In the framework, considering the interaction between errors and deformation, the difference map between geometric errors and pose errors is explicitly and analytically revealed. The mapping matrix can degenerate to be consistent with the previous research under non-over-constrained robots. In addition, the error forward propagation method and the error identification algorithm are proposed to solve the error nonlinear propagation and the ill-posed identification equations, respectively. Finally, the calibration simulation and experiment results of a 5-DOF over-constrained hybrid robot show that the proposed method can accurately predict the pose of the end-effector under geometric errors robustly and effectively.