基于微操作系统的显微立体视觉伺服定位控制研究

发布时间：2018-10-23 21:12

【摘要】：在特征尺寸为微米和亚微米量级的物体上进行加持、吸附、抓取、转移、装配和注射等操作称为微操作。执行微操作的器械设备称为微操作系统或者微操作机器人。近年来,微操作系统被广泛地应用到微机电系统MEMS(Micro Electro Mechanical System)、微光电子机械系统MOEMS(Micro Opto Electro Mechanical System)和生物微机电系统Bio MEMS(Biological Micro Electro Mechanical System)等。将微操作机器人系统和显微视觉信息相结合形成的显微视觉伺服控制系统并使其具有与外部环境进行智能交互能力,是当今微操作机器人系统研究和发展的一个主要方向。目前对微操作系统的研究范围已经从理论研究发展到了应用设计。因此,微操作系统领域的未来发展热点之一是将其大规模的应用到工业生产上。为了实现这一目标,微操作系统自动化和智能化的水平和质量在发展过程中将起到关键的作用。本文围绕微操作系统的自动化问题进行了系统的整体规划,自行构建了一套完整的显微立体视觉伺服微定位控制系统。系统包括:末端执行器模块,运动控制模块和显微视觉模块,并在此基础上对微操作系统中显微立体视觉伺服的成像模型和系统的微定位控制进行了重点研究。针对显微视觉伺服系统中深度信息获取问题,本文基于显微视觉模块建立了G(Greenough)型和CMO(Common Main Objective)型两种结构下的体视显微镜SLM(Stereo Light Microscopy)的成像模型。采用这两种视觉模型可以直接从3D场景中获取物体的视觉信息,避免了微操作系统实时测量或在线估计目标物体未知点的深度,提高了系统的控制性能。在微动机器人运动学的基础上建立了基于图像的视觉伺服控制器,并对微操作闭环系统的稳定性进行了分析。针对显微立体视觉伺服控制系统中稳定性分析复杂的问题,本文建立了基于Hamilton理论的微定位控制算法。通过分解微动机器人的质量矩阵,进行了微操作系统动力学方程的模型变换。进而将显微立体视觉伺服控制系统实现为一类广义Hamilton系统。设计了显微视觉伺服控制器,使得微操作机器人闭环系统渐近稳定。针对传统的基于图像的视觉伺服方法多采用物体的几何特征如点、线、区域面积等作为特征值进行视觉伺服控制时需要图像处理过程中特征值的提取、匹配和跟踪等问题,本文基于Phong照明模式和光流法提出了显微立体视觉系统的微定位方法。此方法是以整幅图像的像素亮度信息作为特征值进行微操作系统的视觉反馈,进而设计了基于亮度的直接伺服控制器,采用此方法避免了图像处理过程中的特征值提取、匹配和跟踪步骤。
[Abstract]:Micromanipulation is called micromanipulation on objects of characteristic size of micron and submicron magnitude, such as adsorption, capture, transfer, assembly and injection. Instruments that perform micromanipulation are called microoperating systems or micromanipulators. In recent years, microoperating systems have been widely used in MEMS (Micro Electro Mechanical System), microelectro-mechanical systems (MOEMS (Micro Opto Electro Mechanical System) and bio-MEMS systems (Bio MEMS (Biological Micro Electro Mechanical System). The micro vision servo control system, which combines the micro manipulation robot system with the micro vision information, has the ability of intelligent interaction with the external environment, which is the main direction of the research and development of the micro operation robot system. At present, the research scope of microoperating system has developed from theoretical research to application design. Therefore, one of the future hotspots in the field of microoperating system is to apply it to industrial production on a large scale. In order to achieve this goal, the level and quality of microoperating system automation and intelligence will play a key role in the development process. In this paper, a complete micro stereo vision servo micro positioning control system is constructed by the overall planning of the system around the automation of the micro operating system. The system includes: end actuator module, motion control module and micro vision module. On the basis of this, the imaging model of micro stereo vision servo and the micro positioning control of the system are studied emphatically. Aiming at the problem of obtaining depth information in micro vision servo system, the imaging model of stereoscopic microscope SLM (Stereo Light Microscopy) with G (Greenough) and CMO (Common Main Objective) structure is established based on microscopic vision module in this paper. By using these two visual models, the visual information of objects can be directly obtained from 3D scenes, which avoids the real-time measurement or on-line estimation of the unknown point depth of the target object by the microoperating system, and improves the control performance of the system. Based on the kinematics of micro-robot, a visual servo controller based on image is established, and the stability of micro-operation closed-loop system is analyzed. Aiming at the complex problem of stability analysis in microstereoscopic vision servo control system, a micro-positioning control algorithm based on Hamilton theory is established in this paper. By decomposing the mass matrix of the micro robot, the model transformation of the dynamic equation of the micro operating system is carried out. Furthermore, the micro stereo vision servo control system is realized as a kind of generalized Hamilton system. A micro-vision servo controller is designed to make the closed-loop system of micro-manipulators asymptotically stable. In traditional image-based visual servo methods, the geometric features of objects such as points, lines, and area areas are often used as feature values for visual servo control, which need to be extracted, matched and tracked in the process of image processing. In this paper, based on Phong illumination mode and optical flow method, a micro positioning method for micro stereo vision system is proposed. In this method, the pixel luminance information of the whole image is used as the eigenvalue for the visual feedback of the micro-operating system, and then a direct servo controller based on brightness is designed, which avoids the extraction of the eigenvalue in the image processing process. Matching and tracking steps.
【学位授予单位】：燕山大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：TP316;TP242

【参考文献】