基于云的语义库设计及机器人语义地图构建

发布时间：2018-02-27 22:28

本文关键词： 智能服务任务语义地图支持向量机多尺度图像分割视差图　出处：《机器人》2016年04期 　论文类型：期刊论文

【摘要】：针对室内移动机器人在智能服务任务中难以获得复杂环境语义的问题,通过设计云端语义库,实现基于语义获取框架的机器人语义地图构建,使机器人不仅掌握面向导航的环境几何描述,而且获得了复杂环境下基于丰富语义库的含物品关联归属关系的语义地图,解决了语义地图构建过程中语义信息添加可靠性低、地图更新存在误差及扩展性不足等问题.首先给出了一种语义库构建方案,基于支持向量机实现语义库分类形成子语义库,在子语义库基础上基于网络文本分类来提取关键特征点形成特征模型库,通过语义分类列表整合子语义库实现物品查询.其次,论述了面向智能服务任务的云端语义地图实现,基于多尺度图像分割与视差图分析,设计标注库与归属库描述物品关联归属关系.最后进行了有关语义地图构建及语义库分类效率的仿真实验与结果分析,验证了方法的有效性.
[Abstract]:Aiming at the problem that it is difficult for indoor mobile robot to obtain complex environment semantics in intelligent service task, the semantic map construction of robot based on semantic acquisition framework is realized by designing cloud semantic library. The robot not only grasps the navigation oriented environment geometry description, but also obtains the semantic map based on the rich semantic library in complex environment, which solves the problem of low reliability of adding semantic information in the process of semantic map construction. There are some problems in map updating, such as error and lack of expansibility. Firstly, a construction scheme of semantic library is presented, which is based on support vector machine (SVM) to realize the classification of semantic database to form sub-semantic library. Based on the sub-semantic database, the key feature points are extracted to form the feature model base, and the item query is realized by integrating the sub-semantic database of semantic classification list. Secondly, the realization of cloud semantic map oriented to intelligent service task is discussed. Based on the multiscale image segmentation and parallax analysis, the article designs the label library and the home library to describe the relationship between the items. Finally, the simulation experiment and result analysis of semantic map construction and classification efficiency of semantic database are carried out. The validity of the method is verified.
【作者单位】：山东大学控制科学与工程学院;山东财经大学计算机科学与技术学院;
【基金】：国家自然科学基金(61573216) 山东省重点研发计划(2015GGX103034) 山东省自然科学基金(ZR2015FM007)
【分类号】：TP242

【相似文献】