摘要
针对即时定位与地图构建(Simultaneous localization and mapping,SLAM)中经典方法的误差累积以及噪声干扰问题,提出基于自适应渐消EKF的SLAM算法。该算法通过引入自适应渐消因子,实时在线调整先验概率密度估计,减小陈旧观测信息对系统估计的影响,在保证协方差矩阵正定性的同时,达到提高SLAM算法估计精度及增强其鲁棒性的目的。通过仿真和基于开源数据集的实验,将提出的算法与EKF-SLAM和UKF-SLAM两种算法进行比较,结果表明AFEKF-SLAM算法在估计精度上优于另外两种算法。
In order to eliminate the error accumulation which inherently exists in the process of Simultaneous localization and mapping(SLAM)and the disturbances of the noise,a SLAM algorithm is proposed involving the adaptive fading extended Kalman filter.By taking advantage of the adaptive fading factor to adjust the prior probability distribution real time online and reduce the effect of old observation information on system estimations,positivity of the covariance matrix is persisted.Besides,the estimation accuracy and the robustness of the proposed algorithm are also improved.The performance of the proposed algorithm is compared with EKF-SLAM and UKF-SLAM under a serial of simulations and experiments occupying an open source data.Results show that the proposed algorithm performs better than the other two algorithms in the estimation accuracy.
作者
杨林志
高宏力
宋兴国
应宏钟
YANG Lin-zhi;GAO Hong-li;SONG Xing-guo;YING Hong-zhong(School of Mechanical Engineering,Southwest Jiaotong University,Sichaun Chengdu610031,China)
出处
《机械设计与制造》
北大核心
2019年第11期249-252,共4页
Machinery Design & Manufacture
基金
国家自然科学基金(51605393)
中央高校基础研究基金(A0920502051619-28)
关键词
移动机器人
即时定位与地图构建
自适应渐消因子
卡尔曼滤波
概率分布
Mobile Robot
Simultaneous Localization and Mapping(SLAM)
Adaptive Fading Factor
Kalman Filter
Probability Distribution