六自由度工业机器人定位误差参数辨识及补偿方法的研究

发布时间：2018-06-10 18:10

本文选题：工业机器人 + 运动学误差　；参考：《华南理工大学》2016年博士论文

【摘要】：现代工业技术的进步促进了机器人性能的不断提高,对工业机器人精度要求也不断提高。特别是离线编程机器人广泛应用,使定位精度问题日益突出。为了提高机器人定位精度,即减小机器人的实际位姿和名义位姿间的误差,利用统计分析方法改进了运动学误差和非几何误差辨识方法,并根据机器人误差源的分类提出了工业机器人误差补偿策略。本文研究了描述机器人末端位姿误差的完整的运动学误差参数选择,并建立运动学误差参数模型。基于Hayati方法修改Khalil的D-H齐次坐标矩阵,建立了各关节的运动学误差参数模型,基于Stone模型建立了基础坐标系和工具坐标系的误差模型。研究了机器人位置误差测量方法、测量坐标系和机器人坐标系的坐标转换方法,提出通过3个位置点测量机器人姿态误差的方法。基于机器人的正运动学和逆运动学方程,提出运动学误差采用正运动学补偿方法,非几何误差采用逆运动学误差补偿方法的策略。研究了两种辨识运动学误差的方法,分别为轴线误差辨识方法和运动学误差参数模型的辨识方法。建立机器人各单关节误差到末端误差的映射,对机器人各关节单轴运动产生的末端误差进行了统计分析,利用统计结果确定了轴线法辨识机器人实际的运动学参数建立坐标系的次序。研究多点拟合空间圆的轴线方向和圆心的方法,根据所获得的各关节轴线特征,依次建立了机器人各关节实际的坐标系,并反向求解获得Hayati修改的D-H模型实际参数的方法。提出了统一的位置和姿态误差辨识运动学误差参数的模型,根据位置误差和位姿误差辨识机器人的运动学误差参数;提出了距离误差辨识运动学误差参数的方法。研究了运动学误差的扩展雅克比矩阵逐项分析线性相关列向量的方法,以及剔除冗余运动学误差参数的方法,改进了从基坐标系到工具坐标系依次剔除冗余误差参数的准则,提出依据末端误差统计分析结果剔除精度高的冗余误差参数的准则,分析得到位姿误差可辨识的28个无冗余运动学误差参数,位置误差可辨识的25个无冗余运动学误差参数。采用矩阵奇异值分解的方法(SVD方法),改进了扩展雅克比矩阵辨识运动学误差参数的计算方法。基于统计方法估算辨识运动学误差需要测量的次数,减少了测量实验次数,保证了扩展雅克比矩阵求解运动学误差参数的计算有效性。根据运动学误差的辨识结果,分别完成机器人的位姿误差补偿、位置误差补偿和距离误差补偿,实验结果表明采用统计分析的方法改进剔除冗余运动学误差参数的方法,可进一步提高机器人的定位精度。利用舍去的二阶运动学误差建立的非几何误差补偿准则,分别计算舍去的二阶运动学误差和非几何误差产生的末端误差并比较两者大小,由此选择需要补偿的非几何误差源。确定补偿机器人连杆自重和机器人末端负载引起的柔性误差,建立了机器人关节的柔性误差到机器人末端误差的映射,基于统计方法分析了各关节的柔性误差对末端误差的影响,确定了需要柔性误差补偿的关节。本文采用3个步骤补偿机器人的柔性误差,辨识了机器人的柔性系数矩阵,在未加负载的情况下补偿了连杆的自重,补偿了由于加载负载产生的柔性误差引起的非几何误差,实验结果表明对统计分析选择的需柔性误差补偿的关节进行补偿,可提高机器人精度和减小外加负载产生的末端误差,并为机械结构直接补偿柔性误差提供研究基础。
[Abstract]:The progress of modern industrial technology has promoted the continuous improvement of robot performance and improved the precision requirements of industrial robots. In particular, off-line programming robots are widely used to make the problem of positioning precision increasingly prominent. In order to improve the positioning accuracy of robots, that is, to reduce the error between the real position and the nominal position of the robot, and to use the statistics. The analysis method improves the kinematic error and the non geometric error identification method, and puts forward the industrial robot error compensation strategy according to the classification of the robot error source. This paper studies the complete kinematic error parameter selection of the robot terminal position and attitude error and establishes the model of the kinematic error parameter. Based on the Hayati method, the Khal is modified. The D-H homogeneous coordinate matrix of IL is used to establish the kinematic error model of each joint. Based on the Stone model, the error model of the basic coordinate system and the tool coordinate system is established. The measurement method of the robot position error, the coordinate system and the coordinate conversion method of the robot coordinate system are studied, and the attitude of the robot is measured by 3 position points. Based on the positive kinematics and inverse kinematics equations of the robot, the kinematic error is proposed by the positive kinematics compensation method, and the non geometric error adopts the strategy of the inverse kinematics error compensation method. Two methods of identifying the kinematic error are studied, which are the identification of the axis error and the identification of the kinematic error parameter model respectively. Method. The mapping of each joint error to the end error of each joint of the robot is set up, the end error of the single axis motion of each joint of the robot is statistically analyzed. The order of the axis method to identify the actual kinematic parameters of the robot is determined by the statistical results. According to the characteristics of each joint axis obtained by the method, the actual coordinate system of each joint of the robot is established in turn, and the method of obtaining the actual parameters of the D-H model modified by Hayati is solved reverse. The model of the unified position and attitude error identification of the kinematic error parameters is put forward, and the kinematics of the robot is identified according to the position error and the position error. The method to identify the kinematic error parameters of the distance error is proposed. The method of analyzing the linear correlation column vector by the extended Jacobian matrix of kinematic error and the method of eliminating the redundant kinematic error parameters are studied, and the criterion of eliminating redundant error parameters from the basic coordinate system to the tool seat system is improved. According to the criterion of eliminating the redundant error parameters with high accuracy according to the statistical analysis results of the end error, 28 non redundant kinematic error parameters can be identified, and 25 non redundant kinematic error parameters can be identified by the position error. The extended Jacobian matrix identification is improved by the method of matrix singular value decomposition (SVD method). The calculation method of the kinematic error parameters. Based on the statistical method, the number of times that the kinematic error needs to be measured is estimated, the times of the measurement are reduced and the calculation validity of the extended Jacobian matrix is guaranteed. According to the identification result of the kinematic error, the position error compensation and the position error of the robot are completed. The result of difference compensation and distance error compensation shows that the method of statistical analysis is used to improve the method of eliminating redundant kinematic error parameters, and the positioning accuracy of the robot can be further improved. By using the non geometric error compensation criterion established by the two order kinematic error, the two order kinematic error and non geometric error of the rounding are calculated respectively. The end error generated by the difference is compared with the size of the two. Therefore, the non geometric error source which needs compensation is chosen. The flexible error caused by the self weight of the connecting rod and the load of the robot end is determined. The mapping of the flexible error of the robot joint to the robot end error is established, and the flexible error pairs of the joints are analyzed based on the statistical method. In this paper, the flexible error of the robot is compensated by 3 steps, and the flexibility coefficient of the robot is compensated. The flexible coefficient matrix of the robot is identified. The weight of the connecting rod is compensated without the load, and the non geometric error caused by the flexible error caused by the load load is compensated. The joint compensation of the flexible error compensation for statistical analysis selection can improve the robot precision and reduce the end error generated by the applied load, and provide the research foundation for the direct compensation of the flexible error for the mechanical structure.
【学位授予单位】：华南理工大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP242

【参考文献】