摘要
信息滤波同时定位与构图(IFSLAM)算法如何获得稀疏化的信息矩阵是解决滤波性能的关键因素。通过对稀疏扩展信息滤波深入分析,发现其稀疏化结果难以保证关联性最弱的环境主动特征被稀疏掉;为了改进这一缺陷,提出一种基于熵稀疏规则的改进SLAM算法,该算法利用熵性质、综合当前以及下一观测时刻来选择与位姿关联性最弱的环境特征作为稀疏特征点;有效提高了算法的稀疏性能。在Matlab上对改进算法进行仿真,验证改进算法的有效性。
How to obtain sparse information matrix is the key factor for information filtering（ IF） simultaneous localization and mapping（ IFSLAM） algorithm to solve filtering performance. By deep analysis on sparse extended information filtering（ SEIF）,finding that it ＇s difficult for sparse result to guarantee the weakest relevance environment active characteristics of being thinning after sparse process; in order to improve this defect,an improved SLAM algorithm based on entropy sparse rules is proposed,it uses the entropy property and considering the current and next observing moment to select posture correlating the weakest environment characteristic as the sparse feature points; effectively improve the sparse performance of the algorithm. The improved algorithm is simulated on Matlab,and validity is verified.
出处
《传感器与微系统》
CSCD
2016年第12期132-136,共5页
Transducer and Microsystem Technologies
基金
浙江省重中之重学科开放基金资助项目(XKXL1514)
浙江省教育厅科研项目(Y201121251)
