SIA OpenIR  > 水下机器人研究室
自主遥控水下机器人水下对接高精度视觉定位方法研究
Alternative TitleResearch on High Precision Visual Positioning Method for ARV Underwater Docking
王丙乾1,2
Department水下机器人研究室
Thesis Advisor唐元贵
Keyword自主遥控水下机器人 视觉定位 水下对接 多目标融合
Pages71页
Degree Discipline机械电子工程
Degree Name硕士
2019-05-17
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract自主遥控水下机器人(Autonomous and Remotely-operated Vehicle, ARV)集自主水下机器人(Autonomous Underwater vehicle,AUV)和遥控水下机器人(Remotely Operated Vehicle,ROV)的部分技术特点于一体,是一种面向极端环境或特殊使命任务的混合型新型水下机器人,具有自主、遥控、混合操作模式。ARV自身携带能源,具有与水面平台相连的微细光纤。利用ARV与水下基站对接可以实现基站与水面平台的实时通信,极大的方便了基站采集的信息的上传与命令的下达,还可以为ARV进行能源补充,加大ARV的持续工作能力,增大其工作范围和工作时长。本文基于水下基站对接口无喇叭状引导机构的实际应用场景,面向对接末端高精度视觉定位需求开展研究。本文通过大量的文献调研,对现有的面向水下机器人与水下基站对接的视觉引导方法进行了总结与分析。结合水下环境从水下有效定位距离、定位精度、计算量、实现成本等方面对现有视觉引导方案进行了评估。现有视觉定位引导方案无法兼顾长可视距离与近距离高精度位姿估计,无法满足对接末端需要高精度定位的需求,本文提出的多目标组合视觉定位方法,能有效兼顾可识别距离远、有效定位范围广、近距离定位精度高、在海流影响下强可靠性等优势。本文利用引导灯、ArUco码、图片特征点三种人工标志组合作为视觉定位目标,形成了远近结合、高低精度互补、可靠性更强的视觉定位方案。主要工作包括:1)在综述了国内外ARV、水下对接、视觉定位等研究成果的基础上,针对ARV水下对接的特点及其对高精度定位的需求进行了高精度视觉定位方法研究。设计了视觉定位算法及视觉引导目标应用方案。2)针对水下对接环境设计了图像处理及目标检测方法。包括图像预处理、引导灯和图片特征点的提取与匹配等。针对水下环境中由于光噪声、弱光照等因素造成的图片特征点误匹配严重以及计算量较大等问题,结合引导灯对ORB特征点提取与匹配进行算法改进。首先通过引导灯提供图片尺度信息,大幅度缩减了计算图片尺度的计算量。此外结合引导灯为ORB特征点设计了相对位子描述子,通过BRIEF描述子与相对位置描述子共同筛选出正确匹配的特征点,能够使特征点匹配的正确匹配率显著提高。3)提出基于多目标的ARV水下单目视觉定位算法。结合ARV水下对接过程及定位算法的性能需求,针对单一定位目标无法满足当前定位需求的现状,提出多目标组合的定位方式,在单种视觉定位目标解算算法的基础上提出了多目标位姿估计结果的融合函数,使得定位算法更加可靠,具有更远的可识别距离和更大的作用区间,近距离具有更高的精度。4)通过陆地及水下实验,对本文改进的特征点提取及匹配算法、多目标组合定位算法进行了验证,并进行了ARV的水池对接实验研究,实现了ARV所搭载的对接机构与对接目标的成功对接,证明了本文所提出的算法在引导ARV水下对接过程中的适用性。
Other AbstractAutonomous and Remotely-operated Vehicle(ARV) integrates some technical features of the Autonomous Underwater vehicle(AUV) and Remotely Operated Vehicle(ROV), is a hybrid new underwater robot for extreme environment or special mission tasks. ARV can worked in autonomous mode, remotely operation mode or hybrid operation mode. ARV itself carries energy and has fine fibers connected to the surface platform. The docking between ARV and the underwater base station realizes real-time communication between the base station and the water surface platform, which greatly facilitates the information that collected by the base station uploading and command release, and can also supplement the energy of the ARV and increase the continuous working capacity of the ARV. In this paper, based on the practical application scenario of the underwater base station without horn-shaped guide mechanism, the requirement of high-precision visual positioning for docking terminal is studied. Through a large number of literature research, this paper summarizes and analyses the existing visual guidance methods for docking underwater vehicles with underwater base stations. Based on the underwater environment, the existing visual guidance schemes are evaluated from the aspects of underwater effective positioning distance, positioning accuracy, calculation coat and implementation cost. The existing visual positioning guidance scheme cannot take into account the long visual distance and the close-range high-precision pose estimation. The proposed multi-target based pose estimation algorithm can effectively take into account the advantages of recognizable distance, wide range of effective positioning, and high precision of close-range positioning and strong reliability under the influence of ocean currents. In this paper, guided lights, ArUco markers and image feature points are used as visual targets to estimate the pose of ARV. Through a certain arrangement, a visual positioning scheme with far and near combination, high and low precision complementarity and more reliability is formed. The main work includes: 1) On the basis of reviewing the research results of ARV, underwater docking and visual positioning at home and abroad, the high-precision visual positioning method is studied for the shortcomings of existing solutions. A visual positioning algorithm and a visual guidance target distribution scheme are designed. 2) Image processing and target detection technology are designed for underwater docking environment. Including image pre-processing, guide light and image feature point extraction and matching. Aiming at the problems of serious mismatching of picture features and large computational complexity caused by light noise and weak illumination in underwater environment, combined with guide light, the algorithm of ORB feature point extraction and matching is improved. First, the picture scale information is provided by the guide light, which greatly reduces the calculation amount of the calculated picture scale. In addition, the relative position descriptor is designed for the ORB feature points in combination with the guide lamp. The correct matching feature points can be selected by the BRIEF descriptor and the relative position descriptor, which can significantly improve the correct matching rate of feature point matching. 3) A multi-target based ARV underwater monocular vision localization algorithm is proposed. Aiming at the current situation that a single positioning target can not meet the current positioning requirements, a multi-target combination positioning method is proposed, and a fusion function of multi-target positioning results is proposed, which makes the positioning algorithm more reliable, with higher precision and larger visible distance interval. .4) The improved feature extraction and matching algorithm and multi-target combination localization algorithm are verified by land and underwater experiments. The applicability of the proposed algorithm in underwater docking environment is proved by underwater docking test.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/25190
Collection水下机器人研究室
Affiliation1.中国科学院沈阳自动化研究所
2.中国科学院大学
Recommended Citation
GB/T 7714
王丙乾. 自主遥控水下机器人水下对接高精度视觉定位方法研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2019.
Files in This Item:
File Name/Size DocType Version Access License
自主遥控水下机器人水下对接高精度视觉定位(3471KB)学位论文 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[王丙乾]'s Articles
Baidu academic
Similar articles in Baidu academic
[王丙乾]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[王丙乾]'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.