提出了一种在非确定环境下求解SLAM数据关联问题的图匹配算法.算法建立了SLAM中数据关联的图论模型,对图模型节点提取了不依赖位置信息的形状上下文特征(shape context,SC),最后通过二次加权随机步进算法(reweighted random walks,RRW)...提出了一种在非确定环境下求解SLAM数据关联问题的图匹配算法.算法建立了SLAM中数据关联的图论模型,对图模型节点提取了不依赖位置信息的形状上下文特征(shape context,SC),最后通过二次加权随机步进算法(reweighted random walks,RRW)得到图匹配问题的优化解.RRW&SC图匹配算法充分利用了路标间的拓扑结构关系以及路标间的形状结构,极大地扩展了数据关联时所依据的几何信息量.仿真实验结果表明,与传统算法相比,该算法能有效处理SLAM中噪声干扰增加、机器人迷失、路标被动态遮挡等不确定程度高、歧义性大环境中的数据关联.展开更多
To tackle the problem of simultaneous localization and mapping(SLAM) in dynamic environments, a novel algorithm using landscape theory of aggregation is presented. By exploiting the coherent explanation how actors for...To tackle the problem of simultaneous localization and mapping(SLAM) in dynamic environments, a novel algorithm using landscape theory of aggregation is presented. By exploiting the coherent explanation how actors form alignments in a game provided by the landscape theory of aggregation, the algorithm is able to explicitly deal with the ever-changing relationship between the static objects and the moving objects without any prior models of the moving objects. The effectiveness of the method has been validated by experiments in two representative dynamic environments: the campus road and the urban road.展开更多
文摘提出了一种在非确定环境下求解SLAM数据关联问题的图匹配算法.算法建立了SLAM中数据关联的图论模型,对图模型节点提取了不依赖位置信息的形状上下文特征(shape context,SC),最后通过二次加权随机步进算法(reweighted random walks,RRW)得到图匹配问题的优化解.RRW&SC图匹配算法充分利用了路标间的拓扑结构关系以及路标间的形状结构,极大地扩展了数据关联时所依据的几何信息量.仿真实验结果表明,与传统算法相比,该算法能有效处理SLAM中噪声干扰增加、机器人迷失、路标被动态遮挡等不确定程度高、歧义性大环境中的数据关联.
基金Project(XK100070532)supported by Beijing Education Committee Cooperation Building Foundation,China
文摘To tackle the problem of simultaneous localization and mapping(SLAM) in dynamic environments, a novel algorithm using landscape theory of aggregation is presented. By exploiting the coherent explanation how actors form alignments in a game provided by the landscape theory of aggregation, the algorithm is able to explicitly deal with the ever-changing relationship between the static objects and the moving objects without any prior models of the moving objects. The effectiveness of the method has been validated by experiments in two representative dynamic environments: the campus road and the urban road.
