基于自适应重采样的同步定位与地图构建

摘要

We studied robot simultaneous localization and mapping. Aiming at the problems of particle degeneration and resampling of the particle set.FastSLAM algorithm based on adaptive resampling was presented. First, the algorithm calculated the number of the effective samples to confirm the degree of particle degeneration,and then set the threshold of the effective samples. When the number of the effective samples was less than the threshold,resampling would be carried out The simulation result showed that, compared with EKF-SLAM, FastSLAM algorithm based on adaptive resampling had higher resampling validity,better robustness and better estimation precise of robot path and the landmark positions estimation.%就移动机器人同步定位与地图构建展开研究,针对FastSLAM算法产生的粒子退化及粒子集重采样问题,提出了基于自适应重采样的FastSLAM算法.该算法首先计算粒子集的有效样本数,确定粒子退化程度.然后设定有效样本阈值,当有效样本数小于阈值时则进行重采样.仿真表明:与EKF-SLAM相比,基于自适应重采样FastSLAM重采样效率更高,鲁棒性更好,在机器人路径和陆标位置的估计上,也具有更高的精度.