基于指数积的youbot机械臂运动学参数辨识研究

发布时间：2018-03-17 01:34

本文选题：运动学标定　切入点：三维扫描仪　出处：《南京邮电大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着科技的进步与信息社会的发展,在航空、医药以及电子产品制造等领域,机器人发挥着重要的作用。在工业生产制造过程中,机器人较高的精度决定了工艺产品的成色。机器人作为生产的主体,受到环境变化、装配形变、齿轮传动、机械磨损等各种因素的影响,导致机器人的内部几何名义参数不能精确地表达机器人执行器末端实际位姿与各个关节之间的关系。因此,需要对机器人的运动学参数进行定期标定。如今,市场上所建立的机器人模型主要运用D-H模型标定方法,虽然此方法较容易实现,但是D-H模型不具备完整性、连续性、参数极小性的特点,极大地影响了机器人作业的性能;同时,在进行运动学参数辨识时,也制约了参数标定方法的实现与应用。本文基于旋量理论,描述机器人运动特性。所采用的指数积建模方法能够很好地克服这些缺点。因此,本文依托南京邮电大学机器人实验室内的Kuka-Youbot机器人作为研究对象,深入研究机器人误差分布问题,并对机器人误差源进行标定与补偿。本文取得的主要成果如下:1.首先,应用李代数、旋量理论描述了机械臂的运动学关系,针对五自由度关节型机器人运动学特性,建立了全局指数积、局部指数积和改进指数积的运动学误差模型。2.针对机器人运动学误差分布,采用全局指数积、局部指数积和改进指数积进行机器人运动学建模。随后基于运动学关系,给出了机器人雅克比矩阵,并且通过仿真软件Matlab进行了数值仿真,分析了一阶线性误差成分以及二阶误差成分的影响。3.基于传统参数标定方法,利用指数积运动学模型,建立了相应的齐次变换误差模型,并且采用空间两点之间的距离误差描述机器人执行器末端的绝对定位精度,给出了基于指数积运动学模型的距离误差标定模型。在辨识过程中发现雅克比矩阵出现奇异现象,采用QR分解以及施密特正交法对冗余参数进行修正,最终实现所有参数的辨识。4.在课题实验部分,测量设备选用天远三维扫描仪。测量精度达到0.02毫米,确保数据采集的精确度,并能对标定前后的机器人末端位姿进行对比分析。在实验的过程中,三维扫描仪单次测量范围较小,需要多次测量拼接数据,才能充分表达机器人末端点相对于机器人基坐标系的空间位置信息,再将采集到机器人位姿信息加入到距离误差模型中进行辨识。最后针对Kuka-Youbot这一型号机器人,应用改进的距离误差方法进行了标定实验,结果显示采用改进的POE标定辨识方法使得机器人末端的距离误差大幅降低,同时机器人绝对位置精度明显提高,标定后的Kuka-Youbot机器人到位精度完全能够到达工业要求。
[Abstract]:With the progress of science and technology and the development of information society, robots play an important role in the fields of aviation, medicine and electronic products. The higher precision of the robot determines the color of the technological product. As the main body of production, the robot is affected by various factors, such as environment change, assembly deformation, gear transmission, mechanical wear and so on. As a result, the internal geometric nominal parameters of the robot cannot accurately express the relationship between the actual position and pose of the end of the robot actuator and the joints. Therefore, the kinematics parameters of the robot need to be calibrated on a regular basis. D-H model calibration method is mainly used in the market. Although this method is easy to realize, the D-H model does not have the characteristics of integrity, continuity and minimal parameters, which greatly affects the performance of robot operation. At the same time, it also restricts the realization and application of parameter calibration method in kinematic parameter identification. Based on spinor theory, this paper describes the kinematic characteristics of robot. The exponential product modeling method used in this paper can overcome these shortcomings very well. Based on the Kuka-Youbot robot in the robotics laboratory of Nanjing University of posts and Telecommunications, the problem of robot error distribution is deeply studied, and the error source of robot is calibrated and compensated. The main achievements of this paper are as follows: 1. The kinematics relationship of the manipulator is described by using lie algebra and spinor theory. The global exponential product is established for the kinematics characteristics of a five-degree-of-freedom robot. The kinematics error model of local exponential product and improved exponential product. 2. According to the kinematics error distribution of robot, global exponential product, local exponential product and improved exponential product are used to model the kinematics of robot. The Jacobian matrix of the robot is given, and the numerical simulation is carried out by the simulation software Matlab. The influence of the first order linear error component and the second order error component is analyzed. 3. Based on the traditional parameter calibration method, the exponential product kinematics model is used. The corresponding homogeneous transformation error model is established, and the distance error between two points in space is used to describe the absolute positioning accuracy of the end of the robot actuator. The distance error calibration model based on exponential product kinematics model is presented. The singularity of Jacobian matrix is found in the identification process. The QR decomposition and Schmitt orthogonal method are used to modify the redundant parameters. Finally, the identification of all the parameters. 4. In the experiment part of the subject, the measuring equipment adopts the Tianyuan 3D scanner. The measuring precision reaches 0.02 mm to ensure the accuracy of the data acquisition. In the process of experiment, the single measurement range of 3D scanner is relatively small, and the splicing data need to be measured several times. In order to fully express the space position information of the end point of the robot relative to the robot base coordinate system, the position and attitude information of the robot is added to the distance error model to identify the robot. Finally, aiming at the model of Kuka-Youbot robot, The calibration experiment is carried out by using the improved distance error method. The results show that the distance error at the end of the robot is greatly reduced by using the improved POE calibration identification method, and the absolute position accuracy of the robot is obviously improved. The precision of the calibrated Kuka-Youbot robot can reach the requirements of industry.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【相似文献】