一种多机器人栅格地图拼接方法的研究  

Research on Grid Map Merging Method of Multi-Robot

作　　者：李雅梅[1] 张瑞 吕猛 LI Ya-mei;ZHANG Rui;LV Meng(Eaculy of Eecrica and Cortral Enginerng Liaoning Technieal Univenity,Huludao 12505,Chima)

机构地区：[1]辽宁工程技术大学电气与控制工程学院,辽宁葫芦岛125105

出　　处：《测控技术》2020年第4期58-61,100,共5页Measurement & Control Technology

摘　　要：多机器人协作式的同步定位与地图创建(SLAM)是自动导航领域的热点研究方向之一,SLAM技术中栅格地图拼接成为重要的技术点。针对采用SIFT算法进行地图拼接栅格的拼接速度较慢、有效匹配点较少、稳定性差等缺点,提出了一种新的栅格地图拼接方法。使用FAST算法快速选定特征点,利用PCA-SIFT算法生成36维特征描述符进行匹配,并借助最大后验一致性算法选取的最佳匹配点进行初始拼接参数计算,使用加权平均算法对待拼接的两幅栅格地图进行拼接融合。使用公开的数据集进行试验,结果表明,所提出的方法稳定性强、拼接速度快,且拼接精度高。Multi-robot’s cooperative of simultaneous localization and mapping( SLAM) is one of the hot research directions in the field of automatic navigation. Grid map merging of SLAM technology become a crucial technology. A new grid map merging method is proposed to solve the problems of slower merging speed,fewer effective matching points,and poor stability in SIFT. FAST algorithm was used to select feature points quickly, PCASIFT algorithm was used to generate 36 dimensions feature descriptors for matching,and the optimal matching points setected by the MAPSAC( maximum a posteriori sample consistency) algorithm were used to calculate the initial merging parameters. The weighted average algorithm fused the two grid maps to be merged. Experiments using public data sets show that the proposed method has strong stability,fast merging speed and high merging accuracy.

关 键 词：多移动机器人系统 栅格地图拼接 最大后验一致性 加权平均算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]