特征序列数据关联机器人同步定位与地图构建

针对噪声不确定性增大的数据关联问题,提出特征点序列数据关联机器人同步定位与地图构建方法。根据机器人环境特征点的空间几何信息,基于图论建立特征点间的信息相关性。利用相邻两步的特征点观测信息协方差的变化,转化成求解特征点TSP问题和特征序列最大相关函数,以此确定所观测特征点的数据关联。实验证明,提出的方法可在噪声不确定性增大的情况下,保证同步定位与地图构建算法的一致性。

Landmark Sequence Data Association Method for Robot Simultaneous Localization and Map Building

For the noise uncertainty increases, a landmark sequence data association (LSDA) method for robot simultaneous localization and map building (SLAM) is proposed. As robot simultaneous localization and map building, the spatial geometry of the landmarks are considered. Then the correlation among landmarks based on graph theory is established. Between the adjacent two-step observations, the difference of innovation covariance is transformed into maximum correlation function of sequence by solving the TSP problem. Then landmark data association is performed. The experiments show that the proposed method can be to ensure the consistency of estimation in the case of uncertainty noise increasing.