基于误差模型与机器学习结合的2PRU-PSR并联机器人误差补偿方法

doi:10.3901/JME.2025.21.192

摘要/Abstract

摘要： 运动学标定是并联机器人精度保障的重要手段。现有的标定方法只能单独对几何误差或非几何误差进行补偿，效率较低，并且未考虑由传统误差模型局限性所导致的部分无法辨识的几何误差影响。针对以上问题，提出了一种基于误差模型与机器学习结合的并联机器人误差补偿方法，并应用于2PRU-PSR并联机器人(P：移动副，R：转动副，U：虎克铰，S：球铰)。首先，基于闭环矢量法建立2PRU-PSR并联机器人的运动学模型，利用矢量微分法建立几何误差模型。其次，应用最小二乘平方和优化函数lsqnonlin对几何参数进行辨识并完成部分几何误差补偿。然后，针对标定后剩余部分无法辨识的几何误差和非几何误差，分别采用BP(Back propagation)神经网络和高斯过程回归(Gaussian process regression, GPR)模型进行误差预测与补偿。最后，通过仿真和实验验证了该方法的正确性和有效性。实验结果表明经过运动学标定并结合BP神经网络预测补偿后2PRU-PSR并联机器人末端平均位置误差由5.121 4 mm减小至0.105 0 mm，平均姿态误差由0.027 6 rad减小至0.003 80 rad；结合GPR模型预测补偿后末端平均位置误差由5.121 4 mm减小至0.091 8 mm，平均姿态误差由0.027 6 rad减小至0.005 68 rad。该方法同样适用于其他并联机器人的误差补偿，以提高末端精度。

关键词: 并联机器人, 几何误差, 运动学标定, 非几何误差, 机器学习

Abstract: Kinematic calibration plays a crucial role in ensuring the accuracy of parallel robots. However, it solely compensates for geometric errors or non-geometric errors without considering the impact of unidentifiable geometric errors caused by the limitations of traditional error models. To address these issues, an error compensation method that combines error model and machine learning for parallel robots is proposed, and it is applied to a 2PRU-PSR parallel robot (where P denotes a prismatic joint, R denotes a revolute joint, U denotes a universal joint, and S denotes a spherical joint). Firstly, the kinematic model of the 2PRU-PSR parallel robot is established based on the closed-loop vector method. The geometric error model is established by the vector differential method. Secondly, the least square optimization function lsqnonlin method is employed to identify the geometric parameters and accomplished some parts of geometric errors compensation. Then, a back propagation (BP) neural network and Gaussian process regression (GPR) model is utilized to predict and compensate for the rest of unidentifiable geometric errors and non-geometric errors. Finally, the proposed method is validated correct and effective through simulation and experimentation. The experimental results indicate that after applying kinematic calibration combined with BP neural network for error compensation, the average position error of the end pose of 2PRU-PSR parallel robot decreases from 5.121 4 mm to 0.105 0 mm, the average orientation error reduces from 0.027 6 rad to 0.003 80 rad; When it combined with the GPR model, the average position error decreases from 5.121 4 mm to 0.0918 mm, the average orientation error reduces from 0.027 6 rad to 0.005 68 rad. This method is also suitable for error compensation of other parallel robots to improve the precision of end effectors.

Key words: parallel robot, geometric error, kinematic calibration, non-geometric error, machine learning

中图分类号:

TP242

柴馨雪, 陶奕良, 祖洪飞, 徐灵敏, 李秦川. 基于误差模型与机器学习结合的2PRU-PSR并联机器人误差补偿方法[J]. 机械工程学报, 2025, 61(21): 192-203.

CHAI Xinxue, TAO Yiliang, ZU Hongfei, XU Lingmin, LI Qinchuan. Error Compensation Method of 2PRU-PSR Parallel Robot Based on Error Model and Machine Learning[J]. Journal of Mechanical Engineering, 2025, 61(21): 192-203.

参考文献

[1] FENG Jin，GAO Feng，ZHAO Xianchao. Calibration of a six-dof parallel manipulator for chromosome dissection[J]. Proceedings of the Institution of Mechanical Engineers Part C Journal of Mechanical Engineering Science，2012，226(4)：1084-1096.
[2] XIAO Pengfei，JU Hehua，LI Qidong，et al. A new fixed axis-invariant based calibration approach to improve absolute positioning accuracy of manipulators[J]. IEEE Access，2020，8：134224-134232.
[3] JOUBAIR A，SLAMANI M，BONEV I A. Kinematic calibration of a 3-DOF planar parallel robot[J]. Industrial Robot，2012，39(4)：392-400.
[4] LUO Ruiqing，GAO Wenbin，HUANG Qi. POE-based parameter calibration for industrial robots considering joint compliance[C]//20215th International Conference on Robotics and Automation Sciences (ICRAS). Wuhan：IEEE，2021：15-22.
[5] HAYATI S，MIRMIRANI M. Improving the absolute positioning accuracy of robot manipulators[J]. Journal of Field Robotics，2010，2(4)：397-413.
[6] LEE R S，LIN Yanhong. Development of universal environment for constructing 5-axis virtual machine tool based on modified D-H notation and openGL[J]. Robotics and Computer Integrated Manufacturing，2010，26(3)：253-262.
[7] 王新杰，范国锋，王才东. 3自由度并联机器人运动学标定方法研究[J]. 机械设计，2017，34(12)：76-81.WANG Xinjie，FAN Guofeng，WANG Caidong. Kinematic calibration method for parallel robot with 3 degree of freedom[J]. Journal of Machine Design，2017，34(12)：76-81.
[8] SUN Siyong，WANG Xin，YANG Xinpeng，et al. Accuracy analysis and calibration of 3-PUS/S parallel battery swapping robot[C]//20144th IEEE International Conference on Information Science and Technology，Shenzhen：IEEE，2014：543-546.
[9] 刘大炜，王立平，李铁民，等. 一类带冗余支链并联机器的运动学自标定[J]. 机械工程学报，2012，48(1)：1-6.LIU Dawei，WANG Liping，LI Tiemin，et al. Self-calibration of a class of parallel kinematic machines with redundant kinematic chain[J]. Journal of Mechanical Engineering，2012，48(1)：1-6.
[10] CHEN I M，YANG Guilin，TAN C T，et al. Local POE model for robot kinematic calibration[J]. Mechanism and Machine Theory，2001，36(11-12)：1215-1239.
[11] HE Ruibo，ZHAO Yingjun，YANG Shunian，et al. Kinematic-parameter identification for serial-robot calibration based on POE formula[J]. IEEE Transactions on Robotics，2010，26(3)：411-423.
[12] SCHRÖER K，ALBRIGHT S L，GRETHLEIN M. Complete，minimal and model-continuous kinematic models for robot calibration[J]. Robotics and Computer-Integrated Manufacturing，1997，13(1)：73-85.
[13] 张俊，蒋舒佳，池长城. 2UPR&2RPS型冗余驱动并联机器人的运动学标定[J]. 机械工程学报，2021，57(15)：62-70.ZHANG Jun，JIANG Shujia，CHI Changcheng. Kinematic calibration of a 2UPR&2RPS redundantly actuated parallel robot[J]. Journal of Mechanical Engineering，2021，57(15)：62-70.
[14] ROTH Z，MOORING B，RAVANI B. An overview of robot calibration[J]. IEEE Journal on Robotics and Automation，1987，3(5)：377-385.
[15] 廖文和，田威，李波，等. 机器人精度补偿技术与应用进展[J]. 航空学报，2022，43(5)：9-30.LIAO Wenhe，TIAN Wei，LI Bo，et al. Error compensation technology and its application progress of an industrial robot[J]. Acta Aeronautica et Astronautica Sinica，2022，43(5)：9-30.
[16] 王谦. 工业机器人运动学参数标定和温度误差补偿研究[D]. 北京：北京工业大学，2017.WANG Qian. Kinematic parameter calibration and temperature error compensation for industrial robots[D]. Beijing：Beijing University of Technology，2017.
[17] ZHOU Jian，NGUYEN H N，KANG H J. Simultaneous identification of joint compliance and kinematic parameters of industrial robots[J]. International Journal of Precision Engineering and Manufacturing，2014，15(11)：2257-2264.
[18] 孙丰磊. 自重构机器人运动学参数标定与非几何误差补偿策略研究[D]. 北京：北京邮电大学，2023.SUN Fenglei. Research on kinematics parameter calibration and non-geometric error compensation of self-reconfigurable robot[D]. Beijing：Beijing University of Posts and Telecommunications，2023.
[19] ANGELIDIS A，VOSNIAKOS G C. Prediction and compensation of relative position error along industrial robot end-effector paths[J]. International Journal of Precision Engineering and Manufacturing，2014，15(1)：63-73.
[20] BAI Ying. On the comparison of model-based and modeless robotic calibration based on a fuzzy interpolation method[J]. International Journal of Advanced Manufacturing Technology，2007，31：1243-1250.
[21] ALICI G，SHIRINZADEH B. A systematic technique to estimate positioning errors for robot accuracy improvement using laser interferometry based sensing[J]. Mechanism and Machine Theory，2005，40(8)：879-906.
[22] YU Dayong. A new pose accuracy compensation method for parallel manipulators based on hybrid artificial neural network[J]. Neural Computing and Applications，2021，33(3)：909-923.
[23] 柴馨雪，李翔毅，汤陈昕，等. 基于共形几何代数的并联机器人逆运动学分析方法[J]. 农业机械学报，2024，55(3)：421-430.CHAI Xinxue，LI Xiangyi，TANG Chenxin，et al. Inverse kinematics analysis method of parallel robot based on conformal geometry algebra[J]. Transactions of the Chinese Society for Agricultural Machinery，2024，55(3)：421-430.
[24] XU Lingmin，YE Wei，LI Qinchuan. Design，analysis，and experiment of a new parallel manipulator with two rotational and one translational motion[J]. Mechanism and Machine Theory，2022，177：105064.
[25] FACHINOTTI V，ANCA A，CARDONA A. A method for the solution of certain problems in least squares[J]. Int. J. Numer Method Biomed Eng.，2011，27(4)：595-607.
[26] LIAO Shuheng，ZENG Qiang，EHMANN K F，et al. Parameter identification and non-parametric calibration of the tri-pyramid robot[J]. IEEE/ASME Transactions on Mechatronics，2020，25(5)：2309-2317.
[27] NGUYEN H N，LE P N，KANG H J. A new calibration method for enhancing robot position accuracy by combining a robot model-based identification approach and an artificial neural network-based error compensation technique[J]. Advances in Mechanical Engineering，2019，11(1)：1-11.
[28] LE P N，KANG H J. Robot manipulator calibration using a model-based identification technique and a neural network with the teaching learning-based optimization[J]. IEEE Access，2020，8：105447-105454.
[29] BAI Ming，ZHANG Minglu，ZHANG He，et al. An error compensation method for surgical robot based on RCM mechanism[J]. IEEE Access，2021，9：140747- 140758.
[30] 黄丽. BP神经网络算法改进及应用研究[D]. 重庆：重庆师范大学，2008.HUANG Li. BP neural network algorithm improvement and application research[D]. Chongqing：Chongqing Normal University，2008.
[31] 冯立颖. 改进的BP神经网络算法及其应用[J]. 计算机仿真，2010，27(12)：172-175.FENG Liying. Improved BP neural network algorithm and its application[J]. Computer Simulation，2010，27(12)：172-175.
[32] ZHANG Dianjin，ZHANG Guangyu，LI Longqiu. Calibration of a six-axis parallel manipulator based on BP neural network[J]. Industrial Robot：The International Journal of Robotics Research and Application，2019，46(5)：692-698.
[33] 孙喜波. BP神经网络算法与其他算法的融合研究及应用[D]. 重庆：重庆大学，2011.SUN Xibo. Study on the fusion of BP neural network algorithm with other algorithms and its application[D]. Chongqing：Chongqing University，2011.
[34] GHANIZADEH A R，HEIDARABADIZADEH N，HERAVI F. Gaussian process regression (Gpr) for auto-estimation of resilient modulus of stabilized base materials[J]. Journal of Soft Computing in Civil Engineering，2021，5(1)：80-94.
[35] LU H，CHEN J，LIAO W. Nonparametric regression via variance-adjusted gradient boosting Gaussian process regression[J]. IEEE Transactions on Knowledge and Data Engineering，2021，33(6)：2669-2679.