Fast map segmentation method based on spectral partition for robot semantic navigation
Fast map segmentation method based on spectral partition for robot semantic navigation
复制标题
DOI：
10.1109/icma.2016.7558709
复制
发表时间：
2016-08
期刊：
影响因子：
--
通讯作者：
Yafu Tian;Ke Wang;Ruifeng Li;Lijun Zhao
中科院分区：
文献类型：
--
作者：
Yafu Tian;Ke Wang;Ruifeng Li;Lijun Zhao
类似于人类感知外界环境的地图分割方法可以降低机器人导航算法和SLAM问题的计算复杂度。介绍了谱聚类方法在地图分割中的应用。然后，本文提出了几种相似性度量准则来构造相似性矩阵。有了这些准则，移动的机器人就可以适应不同的环境。在此基础上，提出了一种基于轮廓系数准则的自适应聚类方法。根据聚类结果，本文提出了一种有效的在线分割方法。最后，在MobileSim平台上进行了仿真实验，验证了该方案的有效性。地图分割是实现信息高内聚、低耦合的方法。分割结果可以大大减少大规模SLAM和导航问题等NP-Hard问题的响应时间。
Map segmentation method similar to the way human percept the external environment can decrease the computational complexity of robot navigation algorithm and SLAM problem. This paper presents an introduction to the application of spectral cluster method in map segmentation process. Then this paper presents several kinds of similarity measurement criteria to construct the similarity matrix. With these criteria, mobile robot can encountor different kinds of environments. Furthermore, this paper presents a self-adaptive clustering method based on silhouette coefficient criteria. As a result of that clustering result, an effective online segmentation method is prepared in this paper. Finally, the results of the experiment simulated on MobileSim platform demonstrate the performance of the proposal. The map segmentation method is which the high cohesion and low coupling are achieved in information. The segmentation results can greatly reduce the response time of such NP-Hard problems as large-scale SLAM and navigation problem.