多自由度机械臂全视场动态测量关键技术研究

This project focuses on the key problems of pose measuring of manipulators with multi-DOF, originally presents a new method for full-field measuring based on a multi-camera dome system in order to meet the high-precision omnidirectional dynamic measuring requirements. Contents of the research include: to build the full-field measuring system model with multi cameras, to determine the physical structure of the full-field measuring system and the geometric constraints between the space orientation of cameras and the effective field of view; to build the coordinate transformation models of the full-field measuring system; to develop multi-scale marker system, and to study marker-based high-precision and high-reliability features recognition method and positioning technique; to study a composite calibration method by combination of the 3D-target calibration and multi-position marker calibration; to optimize the stereo matching algorithm and the pose solving algorithm, and to provide a parallel processing algorithm as well, in order to enhance the system processing speed; to analyze the impact of system structure parameters on measuring errors and quantitatively analyze the accumulated errors of the whole measuring process, and to establish the measuring error model and the error transfer function. The full-field pose measuring method based on a multi-camera dome system for manipulators with multi-DOF, which is proposed in this project aiming at providing an effective full-range and high-precision tracking and monitoring method, is of originality and feasibility. It can also provide a theoretical basis and technical support for full-field measuring in other research and application fields, and therefore has important theoretical significance and application value.

本项目针对多自由度机械臂位姿测量中的关键问题，原创性地提出了基于穹型结构的全视场测量的新方法，利用多摄像机构建全视场测量系统，以满足全方位、高精度的动态多自由度测量要求。研究内容包括：利用多摄像机建立全视场测量系统模型、确定满足全视场测量的物理结构以及摄像机方位与有效视场之间的几何约束关系；建立全视场测量系统下的坐标转换模型；设计多尺度标识系统，并研究基于标识的高精度、高可靠性的特征识别方法与定位技术；研究立体靶标标定与多点特征标识标定相结合的复合标定方法；为提高系统运行速度，优化立体匹配算法、位姿求解算法，设计并行处理算法；分析系统结构参数对测量误差的影响，定量分析测量过程中的累积误差，建立误差模型和误差传递函数。本项目提出的多自由度机械臂全视场测量方法具有独创性和可行性，旨在建立一种有效的全方位高精度的跟踪监测方法，为全视场测量提供理论依据和技术支撑，具有重要的理论意义和应用价值。

为满足全方位、高精度的动态多自由度测量需求，针对多自由度机械臂位姿测量中的关键问题，研究了基于穹顶结构的全视场测量的新方法。在三目视觉的基础上利用多摄像机构建了一种穹顶结构，并基于穹顶结构建立了一种多目视觉全视场测量系统。在此过程中，分析了穹顶结构的空间结构特征，并以此为依据确定了多目视觉中摄像机的空间分布位置和拍摄角度，建立了穹顶结构的空间结构特征与多目视觉可测量区域之间的函数关系，并确定了测量系统的几何约束关系，建立了全视场测量系统下的多自由度坐标系全局转换模型。提取了空间结构特征中的摄像机之间的距离、摄像机与目标物体之间的距离、摄像机拍摄角度等参数，并以此作为特征参数构建了基于穹顶结构的多目视觉模型。针对基于多目视觉全视场测量系统多自由度机械臂位姿测量，设计了多尺度标识系统，并研究了基于标识的高精度、高可靠性的特征识别方法与定位技术，以及立体靶标标定与多点特征标识标定相结合的复合标定方法，并设计了优化立体匹配算法、位姿求解算法和并行处理算法，从而提高了系统运行速度。研究了多目视觉全视场测量系统的结构参数对测量误差的影响，建立了误差模型和测量误差函数，设计了误差补偿策略和方法。.本项目研究的多目视觉全视场测量系统具有独创性和可行性，建立了一种有效的全方位高精度的测量方法，为实现全角度、大范围测量提供了理论支持和技术支撑，具有重要的理论意义和应用价值。