摘要
为了提高六自由度机器人在应用中的定位精度,提出一种提高机器人绝对定位精度的分级标定方法。该方法第一阶段进行几何参数误差的标定,以改进的Denavit-Hartenberg(MD-H)模型为基础,加入减速比和耦合比的因素建立了完整的工业机器人几何参数误差模型,之后采用Levenberg-Marquarelt(LM)算法辨识出机器人的几何参数误差并计算出剩余残差;第二阶段建立基于粒子群—支持向量回归(PSO-SVR)算法的剩余误差预测模型,来预测并补偿修正几何参数后剩余的残留误差。最后,以六自由度工业机器人进行试验验证,经过分级标定后机器人末端中心点的平均位置误差由5.866 mm减少到0.211 6 mm,最大位置误差由10.322 9 mm减少到0.699 9 mm,验证了该标定算法的正确性和有效性。
To improve the positional accuracy of universal robot in the application, a hierarchical calibration method was proposed. The first stage of the method was to calibrate the geometric parameter error. Based on the Modified Denavit-Hartenberg(MD-H) model, a complete industrial robot geometric parameter error model was established by adding the factors of reduction ratio and coupling ratio. After that, the Levenberg-Marquarelt(LM) algorithm was used to identify the geometric parameter errors of the robot and calculate the residuals. In the second stage, a residual error prediction model based on the Particle Swarm Optimization—Support Vector Regression(PSO-SVR) algorithm was established to compensate the residual error after correcting the geometric parameters. The universal six degrees of freedoms Industrial robot was used for experimental verification. After hierarchical calibration, the average position error of the robot end center point was reduced from 5.866 mm to 0.2116 mm, and the maximum position error was reduced from 10.3229 mm to 0.6999 mm, which verified the correctness and effectiveness of the calibration algorithm.
作者
薛祥儒
张承瑞
胡天亮
陈齐志
丁信忠
XUE Xiangru;ZHANG Chengrui;HU Tianliang;CHEN Qizhi;DING Xinzhong(School of Mechanical Engineering,Shandong University,Jinan 250061,China;Shanghai STEP Robotics Co.,Ltd.,Shanghai 201802,China)
出处
《计算机集成制造系统》
EI
CSCD
北大核心
2023年第1期51-60,共10页
Computer Integrated Manufacturing Systems
基金
山东省重大科技创新工程资助项目(2019JZZY020121)
上海市工业强基专项资助项目(GYQJ-2019-1-02)
泰山学者工程专项经费资助项目。
关键词
工业机器人
运动学标定
非几何参数辨识
LM算法
粒子群—支持向量回归算法
industrial robot
kinematics calibration
non-geometric parameter identification
Levenberg-Marquarelt algorithm
particle swarm—support vector regression algorithm
