【作者】 沙梦翎； 王文光；

【Author】 SHA Mengling;WANG Wenguang;School of Electronic and Information Engineering,Beihang University;

【机构】 北京航空航天大学电子信息工程学院；

【摘要】 基于激光雷达的同步定位与地图构建技术具有精度高,受环境因素影响小的特点,但也受累积误差的影响。回环检测技术能够有效减小累计误差,但对场景有一定要求。本文在图优化技术的基础上,给出了一种通过增加边约束改善回环检测效果的方法,包括基于局部边约束的回环检测、基于全局边约束的回环构建和检测。地下车库场景的实验结果表明所提出的改进方法能够增强回环检测的鲁棒性,提高场景重建的准确性。更多还原

【Abstract】 Simultaneous localization and mapping based on lidar is characterized by high accuracy and little influence of environmental factors, but it is suffers from cumulative errors. Based on the graph optimization, this paper presents a novel to improve loopback detection by adding edge constraints, including local edge constraints, and global edge constraints. The experimental results of underground garage show that the proposed method can enhance the robustness of loop detection and improve the accuracy of 3D reconstruction.更多还原