【作者】 李东辉；

【导师】 黄田； 宋轶民；

【作者基本信息】 天津大学 ， 机械制造及其自动化， 2010， 硕士

【摘要】 本文密切结合汽车工业对先进零部件制造装备的重大需求,在国家863计划重点项目课题的资助下,系统研究了新型五坐标可重构混联机器人Tricept-IV的运动学标定方法,内容涉及机构位置分析、误差映射模型、参数辨识、几何误差源分析、误差补偿及相应的计算机仿真等。论文取得以下研究成果:1)利用矢量法建立了含冗余驱动副的5自由度混联机器人Tricept-IV的位置正逆解模型,并借助数值迭代法与全空间比例分配法获得其位置正逆解算法。2)将Tricept-IV并联机构各支链被动关节的姿态误差矢量分解为与位形无关的结构误差和随位形变化的许动方向误差两部分,利用闭环约束方程得到各被动关节许动误差与几何误差参数间的约束关系,进而得到动平台位姿误差与几何误差参数间的误差映射模型,结合串联结构误差传递关系最终构造出Tricept-IV机器人全误差参数映射模型。3)建立了基于激光跟踪仪的误差参数辨识模型,根据误差参数数量级将误差源分为测量坐标系误差、机器人零位误差和几何误差三类,并将其辨识与补偿过程分为三个阶段逐级进行;利用辨识矩阵的最简阶梯行列式的性质将几何误差源分为独立项、相关项和零项三类,并选用了修正系统输入的在线补偿策略。4)以简化误差参数辨识矩阵的条件数为评价指标,综合考虑辨识精度和效率,对测量点位置和数目进行了优化,有效地提高了辨识模型的鲁棒性;计算机仿真结果验证了本文所述运动学标定方法的正确性和有效性。更多还原

【Abstract】 This thesis deals with the kinematic analysis, error modeling, parameter identification, source error analysis, error compensation and computer simulation of a novel reconfigurable hybrid robot—Tricept-IV. The following creative work has been completed.1) The inverse and forward position analysis of Tricept-IV is carried out using vector based method with the consideration of the redundant translational degree of freedom. The corresponding algorithm is programmed using numerical iterative method and proportional distribution principle.2) The orientation error vector of any passive joints in Tricept-IV robot can be decomposed to two parts—constant structural error vector and variable movable error vector. Using the above assumption, the linear mapping between the variable movable error vectors and all of the constant geometrical error parameters has been derived through the closed-loop constraint equations. Applying the linear mapping to one of kinematic chains of the robot, the full geometric error mapping model of the hybrid robot has been obtained.3) The laser tracker is employed to the error measurement and the corresponding parameter identification model is also formulated. According to the size of the source errors, the procedure of the parameter identification and compensation are divided into three steps to identify measurement coordinate system errors, the home position error of the robot and the geometric errors respectively. By analyzing the reduced row echelon form of parameter identification matrix, the source errors are separated to independent, relevant and zero ones. The strategy of the error compensation is carried out by modifying the system input.4) The locations and number of the measurement points is optimized using the condition number of the identification matrix as the evaluation index in order to achieve proper efficiency and robustness. The feasibility and effectiveness of the proposed kinematic calibration method is verified by simulation results.更多还原