室内移动机器人走廊识别及拓扑图构建研究
发布时间：2018-11-03 18:17
【摘要】：近年来,随着移动机器人的应用领域的扩大,越来越多的移动机器人在不同领域为人类服务,代替人类完成复杂和危险的工作任务。移动机器人所处的环境往往是未知、难以预测的。而现有的机器人的智能水平还不能完全满足复杂环境中的工作任务要求,因而提高机器人的智能程度,提高其自主性和适应性,是智能移动机器人发展的关键。本文对智能移动机器人的研究背景做了简要介绍,回顾并分析了机器人室内环境走廊类型识别和拓扑建图的研究现状,指出了目前存在的相关问题,主要研究工作如下：1.对移动机器人室内走廊识别与地图创建研究及发展现状进行研究,综述了机器人在室内走廊识别、环境空间表示及拓扑地图构建方面的研究方法和相关技术。对各种研究方法技术进行了简要的介绍和对比,指出其所具有的特点和存在问题。2.针对室内环境下走廊类型的识别问题,提出了一种改进的走廊识别算法。该算法将走廊分类并编码成八种类型,通过对RGB-D深度传感器提取的环境深度信息数据去噪处理,利用连续点间的深度差值和最小二乘法拟合直线检测走廊出口存在性,最后结合贝叶斯分类器完成对走廊类别的判定。3.研究了室内走廊环境的拓扑图构建问题。根据室内走廊环境的特点,将识别得到的走廊类型作为拓扑图构建中的节点,将走廊过道作为连接节点的分支,创建适用的室内直线式拓扑地图。该拓扑地图结构清晰,简洁,相比于Voronoi图,不容易产生冗余的节点及路径信息。4.研究了多机器人构建的拓扑图合并问题,提出一种使用节点的度和向量匹配算法,提高了顶点的匹配效率。将拓扑地图的合并问题看成是图像匹配与图像配准问题,使用子图匹配的结构融合和图像配准的几何变换等方面的算法进行处理,并采用遍历匹配节点方式,通过寻找来自两个地图之间的最大公共子区域进行地图的合并,最后选出最优匹配方案,完成地图的合并。论文的最后对所做的研究工作加以总结,分析了有待改进的地方,并对未来的发展进行了展望。
[Abstract]:In recent years, with the expansion of the application field of mobile robots, more and more mobile robots serve human beings in different fields, and replace human beings to complete complex and dangerous tasks. The environment in which mobile robots are located is often unknown and unpredictable. However, the intelligent level of the existing robot can not fully meet the requirements of the task in the complex environment, so it is the key to the development of the intelligent mobile robot to improve the degree of intelligence, autonomy and adaptability of the robot. This paper briefly introduces the research background of intelligent mobile robot, reviews and analyzes the research status of indoor environment corridor type identification and topology mapping, points out the existing problems, the main research work is as follows: 1. The research and development status of indoor corridor recognition and map creation of mobile robot are studied. The research methods and related technologies of indoor corridor recognition, environment space representation and topological map construction are summarized. This paper briefly introduces and compares various research methods and technologies, and points out their characteristics and existing problems. 2. An improved corridor recognition algorithm is proposed for the identification of corridor types in indoor environment. The algorithm classifies and encodes corridors into eight types. By denoising the environmental depth information data extracted by RGB-D depth sensor, the depth difference between continuous points and the least square method are used to fit the straight line to detect the existence of corridor exit. Finally, combined with Bayesian classifier to complete the classification of corridor. 3. The problem of building topology diagram of indoor corridor environment is studied. According to the characteristics of the indoor corridor environment, the identified corridor type is regarded as the node in the topology construction, and the corridor is regarded as the branch of the connection node, and the applicable indoor straight-line topological map is created. Compared with Voronoi map, the topology map is not easy to generate redundant node and path information. 4. In this paper, the topological graph merging problem of multi-robot construction is studied, and an algorithm using degree and vector matching of nodes is proposed to improve the efficiency of vertex matching. The merging problem of topological map is regarded as the problem of image matching and image registration. The algorithm of structure fusion of subgraph matching and geometric transformation of image registration is used to deal with the problem, and the method of traversing matching nodes is adopted. By looking for the largest common sub-region between the two maps to merge the map, the optimal matching scheme is selected to complete the map merging. At the end of the paper, the research work is summarized, the improvement is analyzed, and the future development is prospected.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP242
本文编号：2308574
[Abstract]:In recent years, with the expansion of the application field of mobile robots, more and more mobile robots serve human beings in different fields, and replace human beings to complete complex and dangerous tasks. The environment in which mobile robots are located is often unknown and unpredictable. However, the intelligent level of the existing robot can not fully meet the requirements of the task in the complex environment, so it is the key to the development of the intelligent mobile robot to improve the degree of intelligence, autonomy and adaptability of the robot. This paper briefly introduces the research background of intelligent mobile robot, reviews and analyzes the research status of indoor environment corridor type identification and topology mapping, points out the existing problems, the main research work is as follows: 1. The research and development status of indoor corridor recognition and map creation of mobile robot are studied. The research methods and related technologies of indoor corridor recognition, environment space representation and topological map construction are summarized. This paper briefly introduces and compares various research methods and technologies, and points out their characteristics and existing problems. 2. An improved corridor recognition algorithm is proposed for the identification of corridor types in indoor environment. The algorithm classifies and encodes corridors into eight types. By denoising the environmental depth information data extracted by RGB-D depth sensor, the depth difference between continuous points and the least square method are used to fit the straight line to detect the existence of corridor exit. Finally, combined with Bayesian classifier to complete the classification of corridor. 3. The problem of building topology diagram of indoor corridor environment is studied. According to the characteristics of the indoor corridor environment, the identified corridor type is regarded as the node in the topology construction, and the corridor is regarded as the branch of the connection node, and the applicable indoor straight-line topological map is created. Compared with Voronoi map, the topology map is not easy to generate redundant node and path information. 4. In this paper, the topological graph merging problem of multi-robot construction is studied, and an algorithm using degree and vector matching of nodes is proposed to improve the efficiency of vertex matching. The merging problem of topological map is regarded as the problem of image matching and image registration. The algorithm of structure fusion of subgraph matching and geometric transformation of image registration is used to deal with the problem, and the method of traversing matching nodes is adopted. By looking for the largest common sub-region between the two maps to merge the map, the optimal matching scheme is selected to complete the map merging. At the end of the paper, the research work is summarized, the improvement is analyzed, and the future development is prospected.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP242
【参考文献】
相关期刊论文 前3条
1 夏琴晔;杨宜民;;基于biSCAN和SVM的机器人目标识别新算法研究[J];广东工业大学学报;2013年04期
2 蒋玉玲;杨宜民;;基于SOM算法的机器视觉颜色识别[J];广东工业大学学报;2011年02期
3 张晓晖;刘丁;;自主移动机器人走廊视觉识别与跟踪方法研究[J];西安理工大学学报;2006年02期
相关硕士学位论文 前1条
1 王娜;移动机器人拓扑地图创建研究[D];山东大学;2009年
，本文编号：2308574
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2308574.html
