未知复杂环境下多机器人SLAM研究

发布时间：2018-01-02 10:51

本文关键词：未知复杂环境下多机器人SLAM研究　出处：《河北工程大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着机器人相关技术的不断成熟和完善,机器人逐渐的走进了我们的生产和生活中来。智能机器人在未知环境中进行自主导航、编队以及协作任务时,机器人需要获取自身位姿和周围环境信息,从而能够实现避障、路径规划以及编队等。因此作为机器人的关键技术之一的同时定位与地图构建(Simultaneous Localization and Mapping,SLAM)技术仍是当今机器人研究领域的一个重要内容。本文在研究了多机器人系统理论的基础上,结合了国内外多机器人SLAM的相关技术,以快速、高质量的构建环境地图为目的,对多机器人SLAM理论进行了研究。本文的主要研究成果如下:(1)结合现有的实验环境和机器人平台,搭建了机器人位姿模型和环境特征提取模型。为防止环境数据的缺失,设计了基于路径规划的环境数据增补策略,通过获取局部地图的特征点,对特征点进行分析和处理,得到局部地图的剖分三角图,通过对剖分三角图进行路径点的选取,确定机器人环境探测路径,采集不可见区域的数据。采用动态规划目标点和起始点的路径以获取最短路径,降低能量损耗。在Matlab中进行了仿真实验,实验验证了模型的正确性。(2)针对楼宇室内环境中室内结构相似容易造成地图的错误匹配,本文提出一种将Keenan-Motley的室内传播模型和聚类算法相结合的地图融合算法,并构建了地图相关性函数,根据函数值的大小来确定子地图的拼接顺序,防止地图的重复拼接和错误拼接。实验结果表明,该算法能够实现子地图的融合,构建出了全局地图,并且改善了地图的精度,地图的融合效果达到了预期。
[Abstract]:With the robot related technology continues to mature and improve, the robot gradually into our life and production. Intelligent robot navigation in unknown environment, formation and collaborative tasks, the robot needs to get its pose and the surrounding environment information, so as to achieve obstacle avoidance, path planning and formation so as. Localization and map building is one of the key technologies of the robot at the same time (Simultaneous Localization and Mapping, SLAM) is an important content of technology is still the research field of the robot. Based on the research of the multi robot system theory, combined with the related technology at home and abroad, multi robot SLAM to fast, high quality construction environment map to the multi robot SLAM theory are studied. The main results are as follows: (1) combined with the current experimental environment and robot Platform, build a robot pose model and environment model of feature extraction. To prevent the absence of environmental data, design environment data addition strategy based on path planning, the feature points to obtain the local map, analysis and processing of the feature points, get the local map triangulation graph, select path points based on the triangulation graph, determine the robot path, collecting invisible region data. The dynamic path planning objectives and starting points to get the shortest path, reduce the energy loss. In Matlab simulation experiment, the experiment verifies the correctness of the model. (2) according to the indoor environment indoor structure similar mistakes easily lead to map matching, this paper proposed a combination of indoor propagation model and clustering algorithm Keenan-Motley map fusion algorithm, and construct a map correlation function, To determine the order of the stator splicing map according to the function value, to prevent the repetition of stitching and stitching error map. The experimental results show that this algorithm can realize the fusion map, constructed the global map, and improve the precision of map, map fusion achieved the expected results.

【学位授予单位】：河北工程大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【相似文献】