全文下载： 
202306013.pdf
文章编号： 1672-6987（2023）06-0102-08； DOI： 10.16351/j.1672-6987.2023.06.013
马广飞， 邵巍， 李晋屹， 刘祥鹏*(青岛科技大学 自动化与电子工程学院，山东 青岛 266061)
摘要： 针对目前存在的特征点SLAM（同时定位与地图构建）算法在低纹理环境下难以提取足够多的特征点、定位精度低等问题，提出一种基于结构约束的点线特征融合SLAM算法。通过LSD算法提取直线后，将直线分为平行线与垂线，以此作为约束优化特征线的三维位置与相机位姿；通过CAPE算法提取表面法向量估计环境中的曼哈顿主方向，将其作为局部地图的约束优化特征位置与相机位姿。最后，在TUM RGB-D数据集上对提出的算法进行验证。结果表明，该算法提高了在弱纹理环境中的定位精度，数据有效提高了系统的稳定性。
关键词： 同时定位与地图构建； 结构约束； 点线特征融合
中图分类号： TP 242文献标志码： A
引用格式： 马广飞， 邵巍， 李晋屹， 等. 一种基于结构约束的点线特征融合SLAM算法［J］. 青岛科技大学学报(自然科学版), 2023, 44(6): 102-109.
MA Guangfei, SHAO Wei, LI Jinyi, et al. A SLAM algorithm for point-line feature fusion based on structural constraints［J］. Journal of Qingdao University of Science and Technology（Natural Science Edition), 2023， 44（6）： 102-109.
A SLAM Algorithm for Point-Line Feature Fusion Based on Structural Constraints
MA Guangfei, SHAO Wei, LI Jinyi, LIU Xiangpeng
(College of Automation and Electronic Enginnering, Qingdao University of Science and Technology, Qingdao 266061, China)
Abstract: Aiming at the existing problems of SLAM (Simultaneous Localization and Mapping) algorithm of feature points, such as it is difficult to extract enough feature points in low texture environment, low locating accuracy, this paper proposes a SLAM algorithm of point-line feature fusion based on structural constraints. After extracting straight lines by LSD algorithm, the straight lines are divided into parallel lines and vertical lines, which are used as constraints to optimize the three-dimensional position and camera pose of characteristic lines. CAPE algorithm is used to extract the main direction of Manhattan in the surface normal vector estimation environment, which is used as the constraint of local map to optimize the feature position and camera pose. At last, the proposed algorithm is verified on TUM RGB-D dataset. The results show that the algorithm improves the positioning accuracy in weak texture environment and effectively improves the stability of the system.
Key words: simultaneous localization and mapping; structural constraints; point-line feature fusion
收稿日期： 2023-01-31
基金项目： 国家自然科学基金项目(61773227).
作者简介： 马广飞 (1997-)，男，硕士研究生.*通信联系人.
