基于点球约束的机器人误差建模与参数辨识  

Error Modeling and Parameter Identification for Robots Based on Point-sphere Constraints

作　　者：刘毅[1] 金晖力 丰宗强 易旺民 姚建涛[1] 赵永生[1] LIU Yi;JIN Huili;FENG Zongqiang;YI Wangmin;YAO Jiantao;ZHAO Yongsheng(Lab.of Parallel Robotics and Mechatronic Systems in Hebei Province,Yanshan Univ.,Qinhuangdao 066004,China;Inst.of General Assembly and Environmental Eng.,China Academy of Space Technol.,Beijing 100094,China)

机构地区：[1]燕山大学河北省并联机器人与机电系统实验室,河北秦皇岛066004 [2]中国空间技术研究院总装与环境工程研究所,北京100094

出　　处：《工程科学与技术》2023年第6期222-235,共14页Advanced Engineering Sciences

基　　金：国家自然科学基金项目(U2037202,52075466)。

摘　　要：自动化装配对于机器人绝对定位精度提出了更高的要求,机器人理论位姿和实际位姿总存在一定的误差,若绝对定位精度过低,容易导致装配过程中零部件之间发生碰撞,严重影响装配机器人的应用与推广。为此,提出了一种基于点球约束的机器人误差建模与参数识别方法:1)通过在机器人末端安装的6维力传感器反馈末端受力情况,控制机器人以多种姿态使标定锥与靶标球球面重合,记录接触时各关节的位置数据;2)以靶标球球体半径为适应度函数,利用遗传算法辨识误差参数,从而建立完整的误差补偿模型。以自主研制的7自由度装配机器人为研究对象,针对装配机器人的结构特点,由正向递推建立机器人的正运动学方程,应用固定关节法与反变换法获得机器人逆运动学方程;建立机器人的运动学误差模型,预设定误差参数与位姿变换矩阵,通过牛顿迭代法获取了关节变量值,利用遗传算法进辨识误差参数,将辨识结果代入运动学模型中进行验证。采用点球式标定方法采集机器人关节数据,应用遗传算法辨识误差参数,将所得参数代入误差模型中进行实验,结果表明,绝对定位精度提升了76.74%,验证了基于点球约束的机器人误差建模与参数识别方法的有效性,为多自由度机器人标定研究提供了有益参考。Automated assembly puts forward high requirements for the absolute positioning accuracy of the robot.There is always a certain error between the theoretical and actual poses of the robot,and too low absolute positioning accuracy will be likely to cause collisions between parts in the assembly process.This seriously affects the application and promotion of the assembly robot.Therefore,a robot error modeling and parameter identification method based on the point sphere constraint is proposed:1)the robot is controlled to coincide the calibration cone with the sphere of the target sphere in various postures by feeding back the end force condition through the six-dimensional force sensor installed at the end of the robot,and the position data of each joint at the time of contact is recorded;2)the radius of the target sphere is used as the adaptation function,and the error parameters are identified by genetic algorithm,so as to establish a complete error compensation model.The seven-degree-of-freedom assembly robot is used as the research object.The positive kinematic equations of a seven-degree-of-freedom assembly robot are established by forward recursion based on its structural characteristics.The inverse kinematic equations of the robot are obtained by applying the fixed joint method and the inverse transformation method.The joint variable values are substituted into the positive kinematic equations for verification.The genetic algorithm is used to identify the error parameters,and the identification results are substituted into the kinematic model for verification.The experimental data of robot joint are collected by the point-sphere calibration method to identify error parameters through the genetic algorithm.The identified error parameters are substituted into the error model and the calculation results show that the absolute positioning accuracy is increased by 76.74%.This work provides useful references for the positioning calibration of multi-degree of freedom robot.

关 键 词：7自由度机器人 运动学 误差模型 参数辨识 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]