基于因子图优化的水下机器人SLAM算法
SLAM Algorithm of Underwater Robot Based on Factor Map Optimization
谭东旭 1刘鑫宇 1朱杰辉 1韩继群2
作者信息
- 1. 广东智能无人系统研究院(南沙),广州 510000
- 2. 东北大学,沈阳 110000
- 折叠
摘要
提出一种基于因子图优化的多传感器融合水下机器人即时定位与地图构建(Simultaneous Localization And Mapping,SLAM)算法,以解决视觉-即时定位与地图构建(Visual-Simultaneous Localization And Mapping,V-SLAM)在复杂水下环境中面临的问题.该算法结合光纤惯性导航系统、多普勒测速仪和双目相机数据,通过光纤惯性导航系统数据预积分和卡尔曼滤波,提升导航定位精度.利用因子图优化和闭环检测增强系统健壮性和一致性.经过测试,该算法在导航精度和三维建模上优于传统SLAM算法.
Abstract
A Simultaneous Localization And Mapping(SLAM)algorithm for multi-sensor fusion underwater vehicle based on factor map optimization is proposed.To solve the problem of Visual-Simultaneous Localization And Mapping(V-SLAM)in the complex underwater environment.The algorithm combines the data of optical fiber inertial navigation system,Doppler velocimeter and binocular camera,and improves the navigation and positioning accuracy by pre-integrating the data of optical fiber inertial navigation system and Kalman filtering.Factor graph optimization and closed loop detection are used to enhance the robustness and consistency of the system.After testing,the proposed algorithm is superior to the traditional SLAM algorithm in navigation accuracy and 3D modeling.
关键词
水下机器人/因子图优化/即时定位与地图构建(SLAM)算法/多传感器融合Key words
underwater robots/factor graph optimization/Simultaneous Localization And Mapping(SLAM)algorithm/multi-sensor fusion引用本文复制引用
出版年
2024
NETL
NSTL
万方数据