针对6自由度并联平台在圆锥运动轨迹下,几何参数误差和外部测量噪声导致标定精度不足的问题,该文提出了一种基于综合可观测性指标和辨识偏差指标的标定测量位姿联合优化方法。首先,搭建了6自由度并联平台的运动学模型和误差模型,推导了全位姿综合误差Jacobi矩阵;其次,引入全位姿综合可观测性评价指标和辨识偏差评价指标,分别用于表征测量位姿对运动学参数误差的敏感性和对测量噪声的鲁棒性;最后,采用加权多目标遗传算法,在工作空间内选取一组对几何误差敏感、对测量噪声不敏感且具有较好工作空间覆盖性的高质量标定测量位姿。结果表明:相较于随机选取和分级优化方法,该文所提方法运动学参数辨识误差的均方根分别降低了64.74%和55.56%;在圆锥轨迹运动标定后的位置误差和姿态误差方面,该文所提方法均优于对比方法,平均位置误差仅0.003 1 mm,平均姿态误差2.75×10–4°;相较于标定前,标定后平台的平均位置误差降低了84.90%,平均姿态误差降低了69.80%,显著提升了圆锥运动轨迹下的运动精度。该文研究结果可为6自由度并联平台在高精度惯导测试等应用中的运动学标定提供参考。
Abstract
[Objective] Six-degree-of-freedom (6-DOF) parallel platforms are widely used in inertial navigation testing, flight simulation, precision positioning, and robotic manufacturing due to their high stiffness, large load-carrying capacity, and high motion accuracy. Among various motion patterns, conical motion is a typical multi-axis coupled trajectory that continuously excites all axes of motion, making it an important test trajectory for inertial navigation systems and inertial sensors. However, during conical motion, the platform’s motion accuracy is significantly affected by deviations in geometric parameters resulting from manufacturing and assembly errors, as well as measurement noise introduced during calibration. Conventional calibration pose optimization methods mainly evaluate the observability of individual measurement poses, but they neglect measurement noise and the overall distribution of calibration poses, which results in insufficient workspace coverage and poor calibration robustness. [Methods] To address these issues, this study proposes a joint optimization method for calibration measurement poses based on a comprehensive observability index and an identification deviation index. The proposed method improves the accuracy of kinematic parameter identification and enhances robustness against measurement noise, thereby increasing the motion accuracy of a 6-DOF parallel platform under conical motion trajectories. First, the kinematic model and error model of a 6-DOF Stewart parallel platform are established. An error Jacobian matrix containing all geometric parameter errors is derived using the differential method, and a dimensionless normalization scheme is adopted to improve the numerical stability of parameter identification. Second, a comprehensive observability index is proposed based on the global error Jacobian matrix across all calibration poses to evaluate the excitation capability of the calibration pose set with respect to kinematic parameter errors. Third, an identification deviation index is derived from covariance analysis to quantify the influence of measurement noise on parameter identification results, thereby characterizing the robustness of the calibration process under noise. To simultaneously address these two objectives, a weighted multiobjective genetic algorithm is employed to optimize the entire calibration pose set rather than optimizing individual poses as in conventional methods. Based on weight sensitivity analysis, equal weighting is assigned to the comprehensive observability index and the identification deviation index. Considering the 42 kinematic parameters to be identified and the calibration efficiency requirements, 20 calibration poses are selected to construct an overdetermined system of equations, and the kinematic parameters are identified and compensated using an iterative least-squares algorithm. Finally, simulations and experiments are conducted under different conical inclination angles to comprehensively evaluate the calibration performance. [Results] Simulation results demonstrated that, compared with the random pose selection method and the hierarchical optimization method, the proposed method reduced the root mean square error of kinematic parameter identification by 64.74% and 55.56%, respectively, considerably improving parameter identification accuracy. Under conical motion with inclination angles of 5°, 10°, and 15°, the proposed method consistently achieved the highest motion accuracy. In the representative case of a 10° inclination angle, the average position error was only 0.003 1 mm, while the average orientation error reached 2.75×10-4°. Moreover, stable accuracy was maintained even near the workspace boundary. Experimental validation further confirmed the effectiveness of the proposed method. Calibration and compensation decreased the average position error from 0.214 5 mm to 0.032 3 mm, an 84.90% reduction, and decreased the average orientation error from 0.012 9° to 0.003 9°, a 69.80% reduction. Furthermore, calibration considerably suppressed periodic fluctuation in motion errors, indicating that systematic errors were effectively compensated and that the platform’s motion stability was substantially improved . [Conclusions] This study proposes a multiobjective joint optimization method for calibration pose selection of a 6-DOF parallel platform based on a comprehensive observability index and an identification deviation index. By optimizing the entire calibration pose set, the proposed method ensures sufficient excitation of kinematic parameters while improving robustness against measurement noise and maintaining adequate workspace coverage. Simulation and experimental results demonstrate that the proposed method markedly enhances the accuracy of kinematic parameter identification and effectively improves the platform’s position and orientation accuracy under conical motion trajectories. Although this work adopts conical motion as the research scenario, the proposed method is not restricted to a specific trajectory and can be readily extended to other calibration trajectories that satisfy the observability requirements for kinematic parameters. Therefore, it provides a practical and effective solution for the kinematic calibration of high-precision parallel robots and inertial navigation testing systems.
关键词
6自由度并联平台 /
圆锥运动 /
测量位姿优化 /
遗传算法 /
运动学标定
Key words
six-degree-of-freedom parallel platform /
conical motion /
measurement pose optimization /
genetic algorithm /
kinematic calibration
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] TSAI L W. Mechanism design: Enumeration of kinematic structures according to function[M]. New York: CRC Press, 2001.
[2] 李东. Stewart并联机器人的动力学建模与柔顺控制策略[D]. 长春: 吉林大学, 2021. LI D. Dynamic modeling and compliance control strategy of a Stewart parallel robot[D]. Changchun: Jilin University, 2021. (in Chinese)
[3] 陈振宇. Stewart并联机构运动分辨率研究[D]. 成都: 电子科技大学, 2016. CHEN Z Y. Research on motion resolution of Stewart parallel mechanism[D]. Chengdu: University of Electronic Science and Technology of China, 2016. (in Chinese)
[4] WU H Y, KONG L Y, LI Q C, et al. A comparative study on kinematic calibration for a 3-DOF parallel manipulator using the complete-minimal, inverse-kinematic and geometric-constraint error models[J]. Chinese Journal of Mechanical Engineering, 2023, 36(1): 121.
[5] JIANG S J, CHI C C, FANG H L, et al. A minimal-error-model based two-step kinematic calibration methodology for redundantly actuated parallel manipulators: An application to a 3-DOF spindle head[J]. Mechanism and Machine Theory, 2022, 167: 104532.
[6] BORM J H, MENG C H. Determination of optimal measurement configurations for robot calibration based on observability measure [J]. The International Journal of Robotics Research, 1991, 10(1): 51-63.
[7] DRIELS M R, PATHRE U S. Significance of observation strategy on the design of robot calibration experiments[J]. Journal of Robotic Systems, 1990, 7(2): 197-223.
[8] NAHVI A, HOLLERBACH J M. The noise amplification index for optimal pose selection in robot calibration[C]// Proceedings of IEEE International Conference on Robotics and Automation. Minneapolis, USA: IEEE, 1996: 647-654.
[9] WANG W D, SONG H J, YAN Z Y, et al. A universal index and an improved PSO algorithm for optimal pose selection in kinematic calibration of a novel surgical robot[J]. Robotics and Computer- Integrated Manufacturing, 2018, 50: 90-101.
[10] XIANG T W, GAO C H, DU B A, et al. Pose selection based on a hybrid observation index for robotic accuracy improvement[J]. Machines, 2024, 12(8): 501.
[11] CHENG B, WANG B, CHEN S J, et al. A method for calibrating robotic kinematic parameters based on a multi-error source model and an optimized measurement pose set[J]. Industrial Robot, 2025, 52(4): 618-630.
[12] SHIRAFUJI S, GOTO H, ZHANG X T, et al. Visual-biased observability index for camera-based robot calibration[J]. Journal of Mechanisms and Robotics, 2024, 16(5): 051010.
[13] 符磊. 6自由度并联机器人运动学标定的测量构型优化与误差补偿研究[D]. 贵阳: 贵州大学, 2024. FU L. Research on measurement configuration optimization and error compensation for kinematic calibration of 6-DOF parallel robots[D]. Guiyang: Guizhou University, 2024. (in Chinese)
[14] MITCHELL T J. An algorithm for the construction of “D-optimal” experimental designs[J]. Technometrics, 2000, 42(1): 48-54.
[15] 郭万金, 李锦辉, 郝钦磊, 等. 基于病态参数分离的机器人运动学标定测量构型优化[J]. 仪器仪表学报, 2024, 45(2): 299-314. GUO W J, LI J H, HAO Q L, et al. Ill-conditioned parameter separation based optimization of measurement configuration for robot kinematic calibration[J]. Chinese Journal of Scientific Instrument, 2024, 45(2): 299-314. (in Chinese)
[16] JIA H K, ZENG H B, ZHANG J Y, et al. Industrial robot calibration optimization method based on R-optimal criterion and improved IOOPS algorithm[J]. Measurement Science and Technology, 2025, 36(1): 015006.
[17] HUANG C H, XIE F G, LIU X J, et al. Measurement configuration optimization and kinematic calibration of a parallel robot[J]. Journal of Mechanisms and Robotics, 2022, 14(3): 031017.
[18] 余杨, 张洪钺. 圆锥运动及其影响的3种描述方法[J]. 北京航空航天大学学报, 2008, 34(8): 956-960. YU Y, ZHANG H Y. Three descriptions of coning motion and its influence[J]. Journal of Beijing University of Aeronautics and Astronautics, 2008, 34(8): 956-960. (in Chinese)
[19] 杨启航. 基于圆锥运动的MEMS倾角传感器动态测试方法研究[D]. 杭州: 中国计量大学, 2023. YANG Q H. Research on dynamic measurement method of MEMS inclinometers based on conical motions[D]. Hangzhou: China Jiliang University, 2023. (in Chinese)
[20] JIANG X J, FANG L J, GAO Y. A multilevel index optimization method for fast kinematic calibration configuration of serial manipulators based on compute unified device architecture parallel computing[J]. Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science, 2020, 234(23): 4708-4724.
基金
四川省科技计划资助项目(2025YFHZ0030)