中国机械工程 ›› 2024, Vol. 35 ›› Issue (03): 472-480.DOI: 10.3969/j.issn.1004-132X.2024.03.009

• 智能制造 • 上一篇    下一篇

面向物理约束的机器人运动学标定最优位姿集规划方法研究

姜吉光1;侯爵1;苏成志2;巴麒蛟1;田爱鑫1;徐明宇1   

  1. 1.长春理工大学机电工程学院,长春,130022
    2.长春理工大学人工智能学院,长春,130022

  • 出版日期:2024-03-25 发布日期:2024-04-23
  • 作者简介:姜吉光,男,1980 年生,副教授。研究方向为智能装备与制造技术。E-mail:jiangjiguang1980@126.com。
  • 基金资助:
    国防基础科研计划(JCKY2019411B001);吉林省科技发展计划(20210201041GX);吉林省重大科技专项(20220301029GX)

Research on Optimal Pose Set Planning Method under Physical Constraint Robot Kinematics Calibration

JIANG Jiguang1;HOU Jue1;SU Chengzhi2;BA Qijiao1;TIAN Aixin1;XU Mingyu1   

  1. 1.School of Mechanical and Electrical Engineering,Changchun University of Science and
    Technology,Changchun,130022
    2.School of Artificial Intelligence,Changchun University of Science and Technology,
    Changchun,130022

  • Online:2024-03-25 Published:2024-04-23

摘要: 在物理约束下的工业机器人运动学标定过程中,标定精度受到位姿集的影响,而位姿集的选取又受到标定装置的约束,针对以上问题,提出了一种采样区间评价结合位姿集优选的最优位姿集规划方法。首先建立了机器人运动学模型及距离约束标定模型,计算了机器人系统参数误差约束方程及误差雅可比矩阵;然后对机器人工作空间进行空间网格划分,应用拉丁超立方采样结合可观测指标对各个网格区间进行评价,得到最优采样区间;再次基于离线数据建立标定精度预测模型,在最优采样区间内实现最优位姿集的搜索;最后对中瑞RT-608机器人进行最优位姿集的规划及验证,结果表明:基于最优位姿集标定后的平均拟合球半径为0.3947 mm,较随机位姿集减小了57.98%。

关键词: 工业机器人, 运动学标定, 物理约束, 最优位姿集, 区间评价

Abstract: In the processes of kinematics calibration of industrial robots under physical constraints, the calibration accuracy was affected by the selection of the pose set, which in turn was constrained by the calibration devices. To solve these problems, an optimal pose set planning method was proposed based on sampling interval evaluation combined with pose set optimization. Firstly, the robot kinematics model and the distance constraint calibration model were established, and the robot system parameter error constraint equation and error Jacobian matrix were calculated. Secondly, the workspace of robot was divided into spatial grids and evaluate each grid interval using Latin hypercube sampling combined with observable indicators to obtain the optimal sampling interval. Finally, based on offline data, the calibration accuracy prediction model was established based on offline data and search for the optimal pose set within the optimal sampling interval. By planning and verifying the optimal pose set for the ZhongRui RT-608 robot, the results show that the average fitting sphere radius after calibration is 0.3947 mm based on the optimal pose set, which is 57.98% lower than that of the random pose set.

Key words: industrial robot, kinematic calibration, physical constraint, optimal pose set, interval evaluation

中图分类号: