基于地面约束的改进A-LOAM算法  

Improved A-LOAM Algorithm Based on Ground Constraint

作　　者：王乐兵 王挺[1,2] 姜祎[1,2,3] 李亚伟 WANG Le-bingg;ANG Ting;IANG Yi;LI Ya-wei(State Key Laboratory of Robotics,Shenyang Institute of Automation,Chinese Academy of Sciences,Shenyang Liaoning 110016,China;Institutes for Robotics and Intelligent Manufacturing,Chinese Academy of Sciences,Shenyang Liaoning 110169,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]中国科学院沈阳自动化研究所机器人学国家重点实验室,辽宁沈阳110016 [2]中国科学院机器人与智能制造创新研究院,辽宁沈阳110169 [3]中国科学院大学,北京100049

出　　处：《计算机仿真》2023年第5期462-466,共5页Computer Simulation

基　　金：国家自然基金联合基金项目(U20A20201)。

摘　　要：同步定位与地图构建(Simultaneous Localization and Mapping, SLAM)是机器人自主导航的关键技术。针对目前激光SLAM算法用于地面机器人位姿估计时出现的特征匹配可靠性不足、高程误差漂移等问题,提出一种基于地面约束的改进A-LOAM算法。算法首先通过分割地面点优化特征提取的过程从而获取更加可靠的特征点用于帧间匹配,进而在帧图匹配过程提取关键帧构建局部地图并加入地面约束减小高程误差。上述算法还加入了回环检测模块进一步抑制误差累积。实验证明,提出的算法轨迹估计精度高于A-LOAM,且加入的地面约束有效减小了估计误差。Simultaneous localization and mapping(SLAM)is the key technology of robot autonomous navigation.Aiming at the problems of feature matching reliability and elevation error drift when LiDAR SLAM algorithm is applied to ground robot pose estimation,this paper proposes an improved a-loam algorithm based on ground constraint.Firstly,the algorithm optimizes the feature extraction process by segmenting the ground points to obtain more reliable feature points for scan-to-scan matching.Secondly,the algorithm extracts keyframes to construct a local map during scan-to-map matching period,and adds ground constraints to reduce the elevation error.In addition,the algorithm uses loop detection to further suppress error accumulation.Experimental results show that the trajectory estimation accuracy of the proposed algorithm is higher than that of A-LOAM,and the ground constraint can effectively reduce the estimation error.

关 键 词：激光同步定位与地图构建 关键帧 地面约束 闭环检测 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]