CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleVision Measurement and Control for Multi-Part Micro-Assembly Systems
Thesis Advisor徐德
Degree Grantor中国科学院大学
Place of Conferral中国科学院自动化研究所
Degree Discipline控制理论与控制工程
Keyword特征提取 图像雅可比矩阵 视觉伺服控制 显微视觉 微装配 Feature Extraction Image Jacobian Vision Servoing Control Micro-vison Micro-assembly
Abstract由于能够实现高精度的装配操作,微装配技术与系统近年来受到广泛的关注,显微视觉引导下的微装配机器人为微小零件三维空间装配提供了有效的解决方案。本文致力于研究微米级精度的自动微装配系统,以一个由三目显微视觉系统和六个机械手构成的微装配系统为平台,研究基于显微视觉的图像处理和特征提取方法、非正交视觉系统与多个机械臂之间的图像雅可比矩阵标定方法,设计显微视觉/微触觉混合控制策略,实现了高精度的多微装配操作。本论文目前取得了以下研究成果: (1) 针对柱状微零件边缘图像特征,比较基于时域、空域和统计方法的聚焦评价函数,提出了一种粗精结合的微零件特征跟踪自动聚焦算法,实现了对柱状微零件边缘精确的自动聚焦。改进聚焦搜索算法,对聚焦评价函数均值滤波,消除局部极值点,能够有效防止聚焦陷入局部极大值导致的聚焦失败。 (2) 针对圆特征的检测,比较了经典的最小二乘法拟合圆与Hough变换检测圆的方法,提出了一种利用圆弦中垂线拟合圆的方法,可实现检测的高实时性、强抗干扰性和高精确性。 (3) 提出了一种针对多目显微视觉系统的图像雅可比矩阵标定方法,采用主动运动和最小二乘法获取图像雅可比矩阵,该方法能解决非正交视觉系统与任意位置摆放的机械手之间的映射难题;设计了PI控制器,实现了对多个微零件的三维空间位姿的高精度视觉伺服控制。 (4) 为了实现零件间的长距离无损过盈插入,提出基于多目显微视觉系统与微力传感的混合控制方法,采用视觉伺服控制实现零件间的位姿对准,利用力反馈控制器实现零件间的无损插入过程。 (5) 以OpenCV和VS为开发环境,开发了多机械手异型零件微装配平台的控制软件系统,并以该系统为实验平台验证了上述方法。实验结果表明,装配位置误差小于5微米、角度误差小于0.05度,总装配时间小于15分钟。
Other AbstractIn recent years, the research of micro-assembly technology and system have been widespread concerned, because of being able to achieve high precision assembly operations. Micro-assembly robot in the guided of microscope vision provides an effective solution for the three-demensional space assembly of small parts. This topic is dedicated to study the micron grade precision automatic micro-assembly system. On the platform consisting of three vision systems and six manipulators, we study on the micro-vision image processing and feature extraction and recognition methods, and the image Jacobian matrix calibration method between the nonorthogonal visual system and multiple robot manipulators. We have designed the hybrid control strategy of micro-viosn and micro- tactile. Finally multiple micro-assembly operations reach high precision. The main work is described as follows. (1) For the edge characteristics of cylindrical micro-parts, we have compared the focusing evaluation function of time-domain and airspace and statistic methods. In this paper, a coarse-fine autofocus algorithm for tracking micro-parts features has been proposed. Improve focus searching algorithm, employ mean filtering for the focusing evaluation function, and eliminate local extremum points can effectively prevent the focus getting into local maximum value which leads to focusing failure. (2) For the circular feature detection, we have compared the classical least squares fitting circle method and the circle Hough transform detection method. Finally we proposed a highly real-time and strong anti-jamming and highly accurate method using the circular string midperpendicular to fitting the cicle. (3) The image Jacobian matrix calibration method for multiple micro-vison system has been proposed. The image jacobian matrix can be obtained by active movements and the least square method. This calibration method can solve the mapping problem between nonorthogonal visual systems and multiple manipulators in any position. We have designed the PI controller to achieve the visual servoing control of micron grade accuracy for multiple micro-parts three dimension posion and posture alignment. (4) In order to realize the long distance and undamaged interference fit inserting of two micro components, we have proposed hybrid control method based on microscopic visual systems and micro force sensors. Visual servo control is adopted to accomplish posion and posture alignment between the micro-parts, force fe...
Other Identifier201128014628009
Document Type学位论文
Recommended Citation
GB/T 7714
罗李焱. 多零件微装配系统视觉测量与控制技术研究[D]. 中国科学院自动化研究所. 中国科学院大学,2014.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20112801462800(3008KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[罗李焱]'s Articles
Baidu academic
Similar articles in Baidu academic
[罗李焱]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[罗李焱]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.