中国科学院沈阳自动化研究所机构知识库
Advanced  
SIA OpenIR  > 机器人学研究室  > 学位论文
题名: 基于单目视觉的高精度定位及运动估计研究
其他题名: Research on Monocular High-Precision Localization and Motion Estimation
作者: 周船
导师: 谈大龙 ; 朱枫
分类号: TP391.41
关键词: 计算机视觉 ; 位姿估计 ; 摄像机标定 ; 视觉伺服 ; 机器人
索取号: TP391.41/Z73/2005
学位专业: 机械电子工程
学位类别: 博士
答辩日期: 2005-05-25
授予单位: 中国科学院沈阳自动化研究所
学位授予地点: 中国科学院沈阳自动化研究所
作者部门: 机器人学研究室
中文摘要: 基于视觉的机器人定位和运动估计研究是机器人感知和控制的关键技术之一,这项技术利用视觉传感器以被动的方式捕获信息,计算、分析和理解机器人与外部环境间的相对关系,从而为机器人的决策提供支持,能够提高机器人的环境感知能力和局部自主能力。单像机由于配置简单、信息处理快速,而被广泛应用于多种场合,在基于特征的单目视觉定位和运动估计研究中,按照目标物特征信息是否已知可以分成两类:已知模型和未知模型,前者的估计精度较高,一般应用于基于模型的机器人定位中,后者由于其实用性一般用于机器人的视觉伺服控制中。在本论文中,针对这两类问题的现有技术进行分析,并对以下内容进行了深入的研究:(1)在两视点几何关系中,由对极几何约束关系入手,分析已有的基础矩阵描述法和单应矩阵描述法,并推导了两者间的关系;在此基础上,对多图像多特征点过参数情形的单应矩阵估计提出了一种新的基于总体最小二乘和多图像间秩约束的优化求解方法,该方法利用总体最小二乘对多像点的单应矩阵进行优化、用超单应矩阵的秩约束对多图像的单应矩阵进行优化,并进行迭代求解。实验结果表明,该方法的运行结果良好,具有精度高、收敛速度快等优点。(2)在基于已知3目标点模型的位姿估计中,在单目情况下,分析了当摄像机与特征点在空间中处于某种形位时,位姿估计存在唯一解,对特征点和像机的排布、距离、量化误差等因素对位姿估计精度的影响进行了比较;对双像机情形的不确定性进行了分析,明确从两个单目像机中得到的位姿估计结果存在优化的可能,并利用一种新的实时遗传算法实现寻优过程,为了解决传统遗传算法中的早熟和收敛性的问题,同时兼顾并行性和实时性,对适应度函数、编码策略、进化性和方向性算子等方面做了特殊的设计,实验结果表明该方法能够同时满足精度和实时性要求。(3)位姿估计的精度与像机的标定有直接的联系,在单像机的高精度标定中,本论文对各种传统方法进行了详细分析,在此基础上提出了一种新的具有畸变模型的高精度像机标定方法和优化方法,该方法利用5个参数描述镜头畸变,并利用平面单应矩阵进行粗标定,利用遗传算法进行优化,在优化中选用了一种新颖的评价函数,该方法能够实现全部摄像机内外参数的全局优化,实验结果表明:用该方法标定后的摄像机进行位姿估计,其精度较高。(4)在机器人的视觉伺服中,本论文对平面单应矩阵进行了运动分解,并对多解情形的解空间分布及其实际物理意义进行了分析,对于四自由度水下机器人的视觉伺服,提出了用约束运动消除多解情形,同时证明了当摄像机处于某种位形时,存在不可解情形;用一种混合的方法实现六自由度机器人和水下机器人的视觉伺服,仿真实验和真实结果表明该方法具有较好的精度和鲁棒性。
英文摘要: Vision-based robot localization and motion estimation is key problem of robot perception and control. This method computes, analyzes and understands the relative relation between robot and environment in a passive manner. So it provides decision and improves robot’s environment perceptional ability and semi-autonomous ability. Single camera is used in many cases for its simple configuration and rapid processing speed. In the research of monocular robot localization and motion estimation, it can be classified into two ways according to whether the target information is known: model-based and model-unknown. In this thesis, these two kinds of technique are analyzed, and some problems are lucubrated: (1)This thesis analyzes fundamental matrix and homography matrix derived from epipolar geometry, and their relationships are deduced. On the basis of these two transformations, a homography estimation and optimization method is proposed based on Total Least-Square and rank constraint for multi-point and multi-view cases. This technique exploits both TLS for multi-point homography optimization and rank constraint of super-homography for multi-view homography optimi- zation. It uses recursive method to get the solution. Experiment result shows it has good result. (2)In monocular pose estimation, this thesis analyzes that when the camera and control points setup some space configuration there is unique solution. And it compares the precision of pose estimation influenced by different configurations, distances and quantification errors. In binocular case, the uncertainty is analyzed and it is proved that it is possible to optimize the two solutions from two single cameras. Real-time Genetic Algorithm is exploited for global search. In order to overcome pre-mature and convergence problems in traditional Genetic Algorithm, cost function, coding scheme and directional operator are designed specially to guarantee parallelism and real-time performance. Experiment results show this method can satisfy both precision and real-time requirement. (3)It is very important to calibrate camera correctly for precise pose estimation. On the basis of many traditional calibration methods, this thesis proposes a high- precision distorted camera calibration and optimization method for monocular calibration. Five parameters are used to describe lens distortion. And planar homography method is exploited for coarse calibration. In the searching process, a novel cost function and Genetic Algorithm are used to guarantee the global optimization of all the intrinsic and extrinsic parameters. Experiment results indicate that high precision will be obtained if camera, calibrated with this method, is used for pose estimation. (4)In robot visual servoing, planar homography matrix is decomposed. This thesis analyzes the solution space and explains its geometrical meaning. For 4 DOF underwater robot visual servoing, constrained motion is used to eliminate multiple solution cases. It is proved that when camera is setup in particular con- figuration, there are still unsolvable cases. A hybrid control is exploited to implement 6 DOF robot and underwater robot visual servoing. Experiment result shows this technique has good precision and robustness.
语种: 中文
产权排序: 1
内容类型: 学位论文
URI标识: http://ir.sia.cn/handle/173321/9519
Appears in Collections:机器人学研究室_学位论文

Files in This Item:
File Name/ File Size Content Type Version Access License
基于单目视觉的高精度定位及运动估计研究.pdf(2144KB)----限制开放 联系获取全文
Service
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[周船]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[周船]‘s Articles
Related Copyright Policies
Null
Social Bookmarking
Add to CiteULike Add to Connotea Add to Del.icio.us Add to Digg Add to Reddit
所有评论 (0)
暂无评论
 
评注功能仅针对注册用户开放,请您登录
您对该条目有什么异议,请填写以下表单,管理员会尽快联系您。
内 容:
Email:  *
单位:
验证码:   刷新
您在IR的使用过程中有什么好的想法或者建议可以反馈给我们。
标 题:
 *
内 容:
Email:  *
验证码:   刷新

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

 

 

Valid XHTML 1.0!
Copyright © 2007-2016  中国科学院沈阳自动化研究所 - Feedback
Powered by CSpace