基于因子图的多机器人协同算法改进
Improvement of multi-robot cooperative algorithm based on factor graph
秦雨露 1李宏伟 2杨小月 1姜懿芮 1王步云2
作者信息
- 1. 郑州大学计算机与人工智能学院,河南郑州 450001
- 2. 郑州大学地球科学与技术学院,河南郑州 450001
- 折叠
摘要
为解决多机器人协同定位与建图在复杂、大规模场景下耗时久、工作效率低等问题,提出一种融合算法PO-ORB.在因子图模型中引入锚点,用于存储世界坐标系相对位置,将改进后的因子图算法与ORB-SLAM3算法融合,用于帧间优化,结合两种算法的优点,解决大规模问题下多机器人定位精度低和实时性差的问题.通过DBoW2数据库和筛选策略进行地图融合,对全局地图进行优化,提高地图精度.实验结果表明,所提算法能够有效应用于多机器人协同定位与建图.
Abstract
To solve the problem of long time and low efficiency of multi-robot collaborative positioning and mapping in complex and large-scale scenes,a fusion algorithm PO-ORB was proposed.The anchor point was introduced into the factor graph model,which was used to store the relative position of the world coordinate system.The improved factor graph algorithm was fused with the ORB-SLAM3 algorithm,which was used for interframe optimization.Combining the advantages of the two algorithms,the problems of low positioning accuracy and poor real-time performance of multiple robots under large-scale problems were solved.The map fusion was carried out through DBoW2 database and screening strategy,and the global map was optimized to improve the map accuracy.Experimental results show that the proposed algorithm can be effectively applied to multi-robot collaborative localization and mapping.
关键词
同时定位与建图/多机器人协同/因子图/算法融合/地图融合/帧间优化/地图优化Key words
SLAM/multi-robot collaboration/factor diagram/algorithm fusion/map fusion/inter frame optimization/map op-timization引用本文复制引用
基金项目
国家自然科学基金重点基金项目(42130112)
中国工程院专题咨询研究基金项目(HENZT07)
出版年
2024
NETL
NSTL
万方数据