卷积编码器在无人机回环检测上的应用  

Application of Convolution Encoder in UAV Loopback Detection

作　　者：骆旭锋 李智[1] Luo Xu-feng;Li Zhi(School of Electronics and Information Engineering,Sichuan University,Chengdu 610065,Sichuan Province,China)

机构地区：[1]四川大学电子信息学院,四川成都610065

出　　处：《科学与信息化》2022年第7期49-53,共5页Technology and Information

摘　　要：针对VINS-Fusion在无人机同步定位与建图(Simultaneous Localization And Mapping,SLAM)回环检测中由词袋模型所产生的算力消耗较大以及检测表现不佳等问题,文章在VINS-Fusion框架中提出了基于卷积编码器的回环检测改进算法。图像通过卷积编码器得到类方向梯度直方图序列值,通过该值来进行图像的检测匹配。在KITTI数据集上的实验运行表明,该算法在减少了回环检测时间的同时,提高了整个无人机SLAM系统的鲁棒性。Aiming at the problems of large computing power consumption and poor detection performance caused by the bag-of-words model in the simultaneous localization and mapping(SLAM)loopback detection of UAV by VINS-Fusion,the article proposes an improved algorithm for loopback detection in VINS-Fusion framework based on convolutional encoder.The image is passed through the convolutional encoder to obtain the class direction gradient histogram sequence value,and the image detection and matching are carried out through this value.Experimental runs on the KITTI dataset show that the algorithm improves the robustness of the entire UAV SLAM system while reducing the loopback detection time.

关 键 词：无人机定位 同步定位与建图 回环检测 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]