【作者】 王一；

【导师】 叶声华； 邾继贵；

【作者基本信息】 天津大学 ， 测试计量技术及仪器， 2009， 博士

【摘要】 随着汽车制造业的不断发展,基于多关节运动平台的通用工业机器人柔性在线视觉检测技术对车身总成、分总成的质量监测和控制发挥着重大作用。为了保证检测系统的测量精度,必须对系统的各个环节进行标定,其中包含两个最为关键的环节:机器人本体标定,通过对机器人模型参数的准确识别,建立起机器人处于任意姿态时关节变量到末端法兰盘坐标系位姿之间的准确映射关系;现场标定,在复杂的工业现场条件下将各个坐标系统一到全局坐标系下,减小坐标传递过程中的精度损失。论文完成的主要工作有:1将具有视觉功能的工业机器人在柔性制造业中的应用领域进行了概括总结,列举了柔性在线视觉检测系统在国内外的应用现状,并对系统中的关键技术进行了详细的阐述。2对机器人的正运动学模型进行了分析比较。鉴于DH模型存在不足之处,选择MDH模型作为机器人的正运动学模型,在此基础之上推导机器人的定位误差模型和补偿模型。3建立了基于相对位置精度的机器人定位误差补偿模型。利用空间两点的相对位置误差作为修正模型参数的依据,由于表示相对位置的向量包含了3个不同方向的偏差信息,使得标定过程得以简化,更适合在复杂现场条件下采用。4对含有平行四边形机构的机器人进行了结构分析。通过四杆机构的另外一条传递路径确定关节坐标系的变换关系,将该机构的传递误差考虑到机器人的定位误差模型之中,并与不考虑四杆机构的补偿模型进行了比较。5分析了机械臂柔度误差的产生规律。对悬臂梁受应力的变形情况作了理论分析,在外加负载实验基础上建立了机器人柔性关节模型,进一步建立了外加负载柔度误差补偿模型和机械臂自重柔度误差补偿模型。6对机器人运动过程中产生的温度误差进行了有效的补偿。利用多元线性回归方法确定出众多连杆参数中受温度变化影响最为显著的参数,即这些参数对温度误差的产生具有最大贡献,从而建立温度误差补偿模型。7进行了白车身柔性在线视觉检测系统的现场标定并对系统的测量数据进行了精度分析。比较了基于测量姿态和基于机器人正运动学模型的两种全局标定方法,利用基于奇异值分解的坐标配准方法动态地建立车身坐标系,使用碳化硅球体作为测量基准对机器人温度误差进行在线动态补偿。更多还原

【Abstract】 With the development of the automobile manufactory, universal industrial robot flexible and on-line vision inspection systems, which are based on motion mechanism with multi-joint, play an important role in quality monitoring and control of car body assembly and sub assembly. In order to ensure the measurement accuracy, it is necessary to calibrate every part of the system. Two of the most critical steps are calibration of the robot itself, which is to establish the correct mapping from the joint variables to the end-effector at arbitrary posture by exact identification of the model parameters, and on-field calibration, which is to accomplish coordinate systems unification on complex industrial field with small accuracy loss in coordinate transformations. The main work of the thesis is as follows:1. Summarization of the application domains of industrial robots with visual function in flexible manufacturing. Enumeration of the present application situation of flexible and on-line visual inspection systems in the world. Explanation of the common and key technologies exhaustively.2. Analysis and Comparison of different robot forward kinematics. MDH model was adopted to deduce positioning error model and compensation model of the robot because of the drawbacks of DH model.3. Compensation model for robot positioning error which was based on relative position error. Errors in three directions were included in the vector which represents relative position. Therefore, this method was suitable for complex filed with simplified process.4. Analysis of structure of the robot with parallelogram mechanism. The propagation error was taken into consideration in robot positioning error model through the other transformation path of the four-bar. The calibration results were compared with those without four-bar mechanism in compensation model.5. Analysis of regularity of the robot compliance errors. The flexible deformations of cantilever beam imposed on stress were discussed. Flexible joint model was established based on external loads experiments. Compensation models for compliance errors due to external loads and manipulator’s gravity were established.6. Effective compensation for thermal errors due to manipulator motion. Link parameters, which were affected by temperature variation significantly and contribute most to thermal errors, were selected by Multivariate Liner Regression analysis and compensation model for thermal errors was established accordingly.7. On-filed calibration of BIW (Body-In-White) flexible and on-line visual inspection system and accuracy analysis of measurement data. Two global calibration methods which were based on measuring posture and robot forward kinematics respectively were introduced. Car body coordinates were established dynamically by using coordinate registration method based on Singular Value Decomposition. On-line and dynamic compensation for robot thermal errors was implemented by taking spheres which were made of silicon carbide as reference.更多还原