基于机器视觉的移动机器人导航定位系统研究
文献类型：学位论文
| 作者 | 范振 |
| 学位类别 | 硕士 |
| 答辩日期 | 2014-05-28 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 导师 | 张雷 |
| 关键词 | 移动机器人 视觉导航 色带导航 二维码定位 广角摄像机 |
| 其他题名 | Vision Navigation System of Mobile Robot |
| 学位专业 | 模式识别与智能系统 |
| 中文摘要 | 随着信息技术的不断发展，机器人越来越多地应用到人们的日常生活以及实际的生产中，这使得机器人技术的研究与创新也越来越受到人们的重视。移动机器人作为机器人的一种，因为其应用广泛，实用性强等特点已经成为机器人研究的热点之一。随着移动机器人技术的不断成熟，它广泛应用于工业制造、军事、教育、医疗等领域。移动机器人技术是很多技术的融合，例如运动控制技术、避障技术以及导航技术等。移动机器人的导航定位技术是一项基于信息发展而不断创新的技术，近年来随着数字图像处理技术的发展，视觉导航已经成为移动机器人领域研究的热门问题。视觉导航是模仿人的眼睛，通过对采集到图像的识别判断周围环境来确定机器人位置的一种导航方式，这种导航方式具有采集信息量大并且不会对现场环境造成污染的等特点，具有很好的发展前景。视觉导航可分为无地图视觉导航、基于地图的视觉导航以及以及创建地图的视觉导航。考虑到工业现场环境的多变性以及实际工业作业时对导航系统的高性能要求，本文介绍了两种无地图的导航方式：基于色带的移动机器人导航方法以及以及基于二维码的移动机器人定位方法。通过色带导航，移动机器人可以实时纠正其左右位置偏差，保证在行进过程中的位置准确，并结合控制台下达的任务进而完成现场作业。通过二维码定位，可以在工作中阶段性的纠正移动机器人的前后左右位置，并通过二维码角度纠正其行进方向，同时，由于二维码中可以存放很多信息，因此在定位的同时，还可以通过读取存储在二维码中的信息对移动机器人下达任务，进而完成现场作业。本文首先对移动机器人的系统结构做了整体介绍。首先详细说明了现有的典型移动机器人的工作模式以及机器人架构。然后根据本课题介绍了视觉导航传感器的安装方法以及传感器的整体结构和各部分组成，给出了本系统所用到的各部分硬件组成。然后给出了整个视觉导航算法所需要的数字图像处理基础，包含了图像的预处理、图像分割以及在图像特征识别时需要的最小外接矩形获取方法。随之介绍了两种视觉导航算法。在基于色带的导航算法中，详细阐述了色带中心点的获取方法并阐述了分岔路段的判断方法以及双边直线拟合方法。提出了使用快速八邻域轮廓跟踪方法找到连通域轮廓，使用双边界轮廓扫描的方法提取连通域的轮廓信息。随后介绍了基于二维码的移动机器人视觉导航方式。首先探讨了二维码的自身优势以及它的图形结构，然后根据图形结构给出了位置探测图形识别算法以及它的定位方法，通过定位可以实现移动机器人的实时定位，矫正机器人在前进过程中的累积误差；最后介绍了图形旋转方法以及二维码编码解码方法，通过码值的解读，可以很好的向移动机器人下达任务指令，使移动机器人独立的完成现场工作。由于在很多现场环境中传感器安装空间限制，不得不使用广角摄像机来获取更大范围的图像，因此本文提出了广角摄像机在视觉定位中的应用。以广角摄像机像素坐标的反畸变过程为研究对象，提出了一种在畸变对应表的基础上使用中心迭代法以及区域插值法确定物体实际位置的方法。并通过实验证明了该方法能够准确地测量物体的位置，其精度能够满足工业使用要求。最后通过在实际的移动机器人平台上进行实验验证。通过对两种导航方式的验证，证明本文提出的两种视觉导航算法可以很好的满足工业现场对导航系统实时性以及准确性的要求。 |
| 索取号 | TP242.62/F25/2014 |
| 英文摘要 | With the development of information technology, the robot application domain becomes broader and broader. As an improtant type of robots, mobile robot has been widely used in industrial manufacturing,military, education, medical and other fields. As a result, more and more people put their attentions on the research of the innovation of mobile robot. Mobile robot is the integration of many technologies, such as motion control technology, obstacle avoidance and navigation technology. Vision navigation has become much more available for the improvement of mobile robot. It is the imitation of human eyes. With the recognition of the images which are collected by the camera, mobile robot could get its position in real-time and then correct its direction based on the deviation.Vision navigation can be divided into non-map navigation, map-based navigation, and creating map navigation. Considering the industrial environments and the high requirements of the navigation system , this thesis introduces two kinds of non-map navigation : ribbon navigation and two-dimensional code navigation. By ribbon navigation, real-time mobile robot can correct its position deviation which is calculated with the theory position given by the control console, and it could help the mobile robot to finish the task that is assigned by us. By the recognition of the two-dimensional code , the mobile can not only correct the deviation but also could get the task information which is contained in the code. On the other hand, the angle of two-dimensional code could help the robot correct its traveling direction.The first part of this thesis is the introduction of the method the ribbon navigation of mobile robot . Before the state of the navigation algorithm, some entire digital image processing algorithms which are used will be presented. A fast method in the use of eight neighborhood contour tracking is described to get a fast way to find the outline of the connected domain, which is useful in the identification of the ribbon and the two-dimensional code.The second part of this thesis is the introduction of the method of two-dimensional code navigation. The advantages and characteristics of the two-dimensional will be given in this part at first. Based on the structure of the detection pattern in the QRcode, a positioning method is designed to get the location of the QRcode which will help the robot to get its position and guarantee the success of the task. This part also introduces the coding process of QRcode and the decoding method of it. With those methods, we can get the information which will contain the task of the robot and help the robot to finish the working requirement by themselves.The third part of this thesis is the Research on the Using Method of Wide-angle Lens in Vision Localization. In order to ensure the accuracy of wide-angle lens in robot vision localization, the distortion process of wide-angle camera pixel image coordinates and the method to get the ideal and actual world coordinates of the object are discussed in this paper. A new method which use the Center iteration method and the Regional Interpolation method based on the distortion correction table is proposed. With the distortion-free pixel image coordinates, the world coordinates of the object can be obtained. The experimental results show that the method is effective to get the world coordinates of objects with good accuracy for industrial application.The final part of the thesis is the description of the experiment of the navigation system. With the experiment result, we can get the conclusion that the navigation system which is described in this thesis is effective for robots to get the correct position and make sure the success of the field task. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 页码 | 65页 |
| 分类号 | TP242.62 |
| 源URL | [http://ir.sia.ac.cn/handle/173321/14776]  |
| 专题 | 沈阳自动化研究所_其他 |
| 推荐引用方式 GB/T 7714 | 范振. 基于机器视觉的移动机器人导航定位系统研究[D]. 中国科学院沈阳自动化研究所. 2014. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。