【作者】 吴朝明；

【导师】 王代华；

【作者基本信息】 重庆大学 ， 仪器科学与技术， 2013， 博士

【摘要】 微装配技术已成为生产具有复杂三维几何结构和由不同材料制造的微机电系统（Micro-electro-mechanical system，MEMS）的重要手段。受机械手定位精度的限制，微装配需要视觉和力的反馈控制才能实现。然而，在这种通过视觉和力反馈才能实现的微装配中存在的一些问题一直限制了微装配系统的装配精度和装配效率。首先，如何快速获取清晰的多维高分辨率的视觉反馈信息一直是微装配系统中的难点。其次，在微夹钳上集成实现力反馈的微力传感器是困难的。因此，合理配置微装配系统中的显微视觉系统（Microscopic vision system, MVS），提高自动调焦精度和效率和建立有效的微力传感及控制方法是微装配系统能自动、快速和精确地实现装配任务时必须解决的问题，开展这方面的研究具有重要学术意义和工程应用前景。为解决上述问题，根据建立的显微视觉系统的视觉可分辨能力模型和视场空间模型，提出了一种用于权衡双目显微视觉系统的视觉可分辨能力和视场空间的显微视觉系统Pareto优化方法；通过将调焦评价函数曲线以峰值为界分为左右两条独立的调焦评价函数曲线及将最佳聚焦位置看做这两条独立的调焦评价函数曲线的交点，提出了一种基于双侧预测求交的自动调焦方法；建立了夹持力与微夹钳夹爪角度变化量正切之间的关系，提出了夹持力的显微视觉伺服控制方法。本文的主要研究工作和创新点可以归纳为以下五个方面：1．建立了一个单操作手5自由度自动微装配系统，为自动微装配系统的关键技术研究提供了技术支撑和实验条件。分析了压电致动微夹钳的位移和夹持力特性，建立了夹持力与压电致动微夹钳夹爪角度变化量正切的关系。2．为计算双目显微视觉系统的视场空间的相交体积，建立了双目显微视觉系统的相交视场空间模型。提出了一种显微视觉系统的Pareto优化方法权衡自动微装配系统中的双目显微视觉系统的视觉可分辨能力和相交视场空间，优化双目显微视觉系统的拓扑关系和光学放大倍数。通过在视觉可分辨能力模型和相交视场空间模型基础上建立的多目标函数和“Pareto优化”原理，获得表征权衡双目显微视觉系统的视觉可分辨能力性能和相交视场空间性能之间的关系的Pareto最优曲线（即Pareto前沿）。根据Pareto前沿，双目显微视觉系统的拓扑关系和光学放大倍数能够被确定。3．为实现自动微装配过程中的自动调焦，提出并研究了一种双侧预测求交自动调焦方法。该方法将调焦评价函数曲线以峰值为界分为左右两条独立的调焦评价函数曲线，最佳聚焦位置看作是这两条独立的调焦评价函数曲线的交点。这种自动调焦方法的调焦精度和调焦效率被测试并与7点爬山法和双曲线拟合方法的调焦精度和调焦效率进行了比较。4．为避免在夹持过程中损伤或丢失零件，根据建立的夹持力与微夹钳夹爪角度变化量正切之间的关系，提出了夹持力显微视觉伺服控制方法。应用基于轮廓点余弦值的角点提取方法，根据设定的轮廓点余弦值阈值，提取零件轮廓上的角点和融合图像中零件之间的分界点。研究了基于图像的显微视觉伺服控制方法，设计了基于图像的显微视觉伺服控制的PID（Proportion Integration Differentiation）控制器。在建立的自动微装配系统上，用零件和夹爪位置及夹爪间距控制实验测试了基于图像的显微视觉伺服控制方法，用夹持力的控制实验测试了夹持力显微视觉控制方法。5.为实现微型金属圆柱腔的自动对准和堆叠，规划了这个微装配任务的装配序列。在构建的自动微装配系统上，测试并完成了微型金属圆柱腔的自动对准和堆叠任务。本文的研究工作对自动微装配系统的开发和设计具有重要的指导意义。更多还原

【Abstract】 Microassembly has become an important means that produces themicro-electro-mechanical system (MEMS) with complex three-dimensional geometricstructures and different materials. The microassembly only can be implemented byvision and force feedback control due to the restriction of the positioning accuracy ofthe micromanipulators. However, the assembly accuracy and efficiency of themicroassembly systems have been influenced by the vision and force feedback. Firstly,how to quickly get the legible multidimensional vision feedback information with highresolution is always the issue of the microassembly system. Secondly, it is of greatdifficulty to integrate the micro force sensor on the microgripper. Therefore, designingthe proper microscopic vision system of the microassembly systems, improving theaccuracy and efficiency of the autofocus, and establishing the effective micro-forcecontrol method are necessary to automatically, rapidly, and accurately implementmicroassembly tasks. These researches are of importance due to their academicsignificance and engineering application prospect.In order to solve the above problems, the Pareto optimization method of themicroscopic vision systems for tradeoff between the vision resolvability and theintersecting space of view of the binocular microscope vision system is proposedaccording to the vision resolvability model and the established space of view model inthis study; the bilateral prediction and intersection calculation autofocus method isproposed by dividing the focus measure curve into two independent focus measurecurves with the peak and regarding the intersection of the two independent focusmeasure curves as the best in-focus position; the relationship between the holding forceand the angle change tangent value of the gripping jaws of the microgripper isestablished and the microscopic visual servo control based holding force control methodis proposed as well.The major research works and the innovations in this dissertation can besummarized as follows:1. The automated microassembly system with single manipulator and five degreeof freedom is set up for providing the technical support and the experimental conditionswhen studying these key technologies about the automated microassembly systems. Thedisplacement and force features of the piezoelectric-driven microgripper are analyzed and the relationship between the holding force and the angle change tangent value of thegripping jaws of the piezoelectric-driven microgripper is established.2. The model of the intersecting space of view of the binocular microscope visionsystem is set up for calculating the intersecting volume of their space of views. Themicroscopic visual servo control based Pareto optimization method is proposed foroptimizing the topological relationship and the magnification by weighing the visionresolvability and the intersecting space of view of the binocular microscope visionsystems. The Pareto optimal curve (namely Pareto front) that represents the weighedrelationship between the vision resolvability and the intersecting space of view iscalculated by the multi-objective function established on the vision resolvability and theintersecting space of view and the "Pareto optimization" principle. The topologicalrelationships and the magnifications of the binocular microscope vision system can bedetermined by the Pareto front.3. The bilateral prediction and intersection calculation autofocus method isproposed for implementing the autofocus in the process of the microassembly. Theauto-focus method divides the focus measure curve into two independent focus measurecurves with the peak and regards the intersection of the two independent focus measurecurves as the best in-focus position. The theoretical analyses have shown that theauto-focus method cannot only effectively avoid the principle error caused by assumingthe symmetrical focus measure curve in the auto-focus methods based on curve fitting,but also eliminate the possible waver search near the peak position in the modified fastclimbing search method. The focusing accuracy and the focusing speed of the autofocusmethod have been tested and compared with those of the7-point hill-climbing searchmethod and the quadratic curve fitting method.4. The microscopic visual servo control based holding force control method isproposed according to the established relationship between the holding force and theangle change tangent value of the gripping jaw of the microgripper. The corner pointextraction method based on the cosine value of the contour point can accurately extractthe corner point of the contour of the part and the demarcation point between thedifferent parts in the fusion image if the threshold of the cosine value of the contourpoint can be properly set. The vision-based microscopic visual servo control method isexplored and the PID (Proportion Integration Differentiation) controller of thevision-based microscopic visual servo control method is designed. In the builtautomated microassembly system, the vision-based microscopic visual servo control method is tested by the controlling the part and microgripper positions and the gapbetween the gripping jaws, and the microscopic visual servo control based holding forcecontrol method is also by controlling the holding force when gripping the part.5. For implementing the automatic alignment and stack task of the tiny metalcylindrical cavities, the assembly sequence of the microassembly task is planned. Onthe built automated microassembly system, the automatic alignment and stack task ofthe tiny metal cylindrical cavities is tested.The researches in this dissertation would be beneficial to developing and designingthe automated microassembly systems.更多还原