摘要
同步定位与地图构建(SLAM)是实现移动机器人自主导航定位的关键。针对室内大型环境下激光SLAM闭环检测容易产生错误闭环,导致机器人位姿估计误差较大的问题,提出了一种融合Wi-Fi与激光信息的图优化SLAM算法。首先,构建Wi-Fi指纹序列与激光子地图;然后,根据每对指纹序列的相似度均值和标准差筛选用于闭环检测的激光子地图。在此基础上,提取所筛选子地图的特征点并匹配,以确定激光闭环;最后,通过图优化方法融合里程计与激光闭环,优化机器人的轨迹并构建全局地图。在170 m×30 m和180 m×80 m的室内环境中采集了三组数据,对所提算法性能进行验证。实验结果显示,所提算法的定位精度在三组数据上分别达到0.78 m、0.67 m和0.89 m,与激光SLAM算法相比分别提升了48.6%、53.1%和68.7%,证明所提算法有效提高了室内大型环境下激光SLAM的位姿估计精度。
SLAM is crucial for the autonomous navigation and positioning of robots.Aiming at the problem of considerable pose estimation errors for robots,which were caused by incorrect loop closures in LiDAR SLAM within large-scale indoor environments,this paper proposed a graph-based SLAM algorithm that fused Wi-Fi and LiDAR information.Initially,the algorithm constructed Wi-Fi fingerprint sequences and LiDAR submaps.Subsequently,it selected LiDAR submaps for loop closure detection based on the mean and standard deviation of similarity between each pair of fingerprint sequences.Then,it extracted feature points from the selected submaps and matched them to confirm LiDAR loop closure.Ultimately,using a graph optimization approach,odometry and LiDAR loop closures were fused to optimize the robot’s trajectory and construct a global map.Three datasets were collected in 170 m×30 m and 180 m×80 m indoor environments to verify the performance of the proposed algorithm.The experimental results show that positioning accuracy values of proposed algorithm in three datasets reach 0.78 m,0.67 m,and 0.89 m,which give improvements of 48.6%,53.1%,and 68.7%when compared to the LiDAR SLAM algorithm,demonstrating its effectiveness for enhancing pose estimation accuracy in large-scale indoor environments.
作者
熊壮
刘冉
郭林
肖宇峰
Xiong Zhuang;Liu Ran;Guo Lin;Xiao Yufeng(School of Information Engineering,Southwest University of Science&Technology,Mianyang Sichuan 621000,China;Robot Technology Used for Special Environment Key Laboratory of Sichuan Province,Southwest University of Science&Technology,Mianyang Sichuan 621000,China)
出处
《计算机应用研究》
北大核心
2025年第3期812-817,共6页
Application Research of Computers
基金
四川省科技计划资助项目(2023NSFSC0505,2022YFG0242)
国家自然科学基金资助项目(12175187,12205245)。
关键词
Wi-Fi指纹序列
激光子地图筛选
闭环检测
图优化
同步定位与地图构建
Wi-Fi fingerprint sequence
LiDAR submap selection
loop closure detection
graph optimization
simulta-neous localization and mapping(SLAM)
作者简介
熊壮(1999—),男,四川遂宁人,硕士研究生,主要研究方向为Wi-Fi定位、SLAM;通信作者:刘冉(1986—),男,安徽淮北人,副研究员,硕导,博士,主要研究方向为机器人导航定位、室内定位、SLAM(ran.liu86@hotmail.com);郭林(1999—),男,四川资阳人,硕士研究生,主要研究方向为人机交互、室内定位;肖宇峰(1978—),男,湖南常德人,教授,博导,博士,主要研究方向为网络通信系统、智能机器人系统.
