改进的视觉同时定位与地图创建数据关联方法

Improved Data Association Method of Vision-based Simultaneous Localization and Mapping

数据关联的规模随地图的不断增长而增加是导致机器人同时定位与地图创建(simultane-ous localization and mapping,SLAM)实时性能差的一个主要原因,极小连通支配集(connected domi-nating set,CDS)方法降低地图元素的数目,从而提高数据关联的计算效率,为了进一步优化MCDS方法的性能,对它进行了两处改进:一是延迟建立极小连通支配集;二是自适应地搜索极小连通支配集。在视觉SLAM中,应用SIFT算法提取自然路标,扩展卡尔曼滤波算法融合视觉信息与机器人位姿信息完成SLAM任务。实验结果表明,改进的极小连通支配集数据关联结果是可信的,减少SLAM计算复杂度的性能突出。

The scale of data association increases with the map grows,which is one of the major reasons of poor real-time performance of robot in process of Simultaneous Localization and Mapping（SLAM）.The connected dominating set（CDS） approach is used to reduce the number of landmarks that need to be maintained in the map,which improve the computation efficiency of the data association.Two improvements are introduced to improve the CDS′S performance.Firstly,CDS is constructed lingeringly.Secondly,CDS is searched adaptively.In vision based SLAM,SIFT（Scale Invariant Feature Transform） algorithm is used to extract the Natural landmarks;SLAM is completed by fusing the vision information and robot pose with Extended Kalman Filter（EKF） indoor environment.Experiment results indicate that improved connected dominating set data association results are reliable;the capability of reducing computational complexity is outstanding.