CASIA OpenIR  > 毕业生  > 硕士学位论文
Thesis Advisor易建强
Degree Grantor中国科学院研究生院
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机器人 视觉 Mobile Robot Localization Object Recognize Computer Vision
Abstract随着机器人技术的不断发展,机器人的应用领域不断扩大。人们希望机器人不仅能够在生产车间环境下完成大规模的装配任务,而且也能在室内和室外环境下完成比较复杂的任务。上述任务的完成都需要比较精确地实现机器人的定位。视觉传感器因其所含的丰富信息,也受到了越来越多的关注,研究基于视觉的移动机器人定位具有非常重要的理论意义和较大的实用价值。本文以实验室自主移动机器人为试验平台,以完成机器人在室内环境下比较准确的自定位任务为主线,对机器人定位方法的实现和定位精度的提高方法进行了研究。 第一,对在室内环境下自主移动机器人的定位技术、计算机视觉技术以及机器人的视觉系统等问题进行综述和分析,并对本文的选题背景、主要内容作了介绍。 第二,建立目标物体的颜色模型和形状模型,并用于室内环境下物体的识别。首先深入分析各种颜色空间的适用性和优缺点,选择最合适的颜色空间,并在此空间中利用统计学习的方法建立了目标物体的颜色模型。根据目标物体的形状特征,建立了物体的形状模型。采用颜色模型和形状模型相互结合的方法,使机器人能够迅速、准确地识别室内环境下的目标物体。 第三,给出在实际中运用直接线性变换(DLT)方法来实现摄像机标定的过程,并对标定的结果进行了分析。针对摄像机云台放置不平以及云台控制精度不高对摄像机的外部参数所产生的影响,采用了多次标定的方法,得到了摄像机姿态随云台运动角度之间的变化关系。利用此关系对定位结果进行校正,极大地减小了上述因素对机器人的定位产生的影响。 第四,实现一种基于点对的视觉定位算法,并分析了这种算法的鲁棒性。给出了一种在目标物体上比较精确的提取特征点的方法,针对定位精度不高的问题,提出了一种基于离线学习的方法来提高视觉定位精度的方法,并给出了相关的试验结果。 最后,本文对所取得的研究成果进行了总结,并指出后续需要开展的工作。
Other AbstractThis thesis focuses on the realization of vision based mobile robot localization in indoor environment and the method to improve the precision of the localization. Firstly, computer vision and robot vision system are analyzed. The main research content and development directions of the localization for mobile robots are summarized. The background and the structures of this thesis are also introduced. Secondly, this paper build color model and frame model of target objects, and are analyzed and the color space that is most suitable for this application is chosen. Then, the color model of the object is built using the statistical learning method, which is used together with the frame model of the object to realize the target objects quickly and exactly. Thirdly, the process of camera calibration using the Direct Linear Transformation (DLT) method is given in this paper, and the results of the camera calibration is analyzed. To reduce the influence of unevenly placed platform and the low control precision of the platform, this paper adopts a method of numerous calibrations to get the relationship between the camera posture and the angle of the platform movement. Utilizing this relationship to correct the result of localization can reduce the influence greatly. Fourthly, This thesis introduces a vision based localization algorithm, which only uses two points on the same level, and analyzes the robustness of this algorithm. At the same time, a method to abstract the characteristic points exactly is prompted. To solve the problem of low precision of the robot localization, which may not meet the requirement for precisely positional information in certain tasks, this paper presents a method based on off-line learning to improve the precision of robot localization. Finally, the obtained research results are summarized and future work is addressed.
Other Identifier200228014603513
Document Type学位论文
Recommended Citation
GB/T 7714
李新征. 基于视觉的机器人定位方法及其实现[D]. 中国科学院自动化研究所. 中国科学院研究生院,2005.
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[李新征]'s Articles
Baidu academic
Similar articles in Baidu academic
[李新征]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[李新征]'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.