Computer Engineering and Applications ›› 2023, Vol. 59 ›› Issue (3): 264-275.DOI: 10.3778/j.issn.1002-8331.2108-0304

• Engineering and Applications • Previous Articles     Next Articles

Global Localization Method Based on Global Feature Point Matching

PENG Zexin, ZENG Bi, LIU Jianqi   

  1. School of Computers, Guangdong University of Technology, Guangzhou 510006, China
  • Online:2023-02-01 Published:2023-02-01

基于全局特征点匹配的全局定位方法

彭泽鑫,曾碧,刘建圻   

  1. 广东工业大学 计算机学院,广州 510006

Abstract: To solve the problem of high demand of sensor and large amount of calculation in traditional global localization method, this paper proposes a novel localization algorithm for mobile robots based on global feature point matching using 2D LiDAR sensors. The global feature points are synchronously extracted while global map building. Within the global localization algorithm, local maps are established and local feature points are further extracted, then the global pose can be obtained by a real-time comparison and matching between local and global feature points. The proposed method performs better than the AMCL(adaptive Monte Carlo localization) and Cartographer methods in positioning accuracy and calculating speed on two test sets. The results shows that, this method can effectively make the robot localization faster, as well as reduce the consumption of computing resources.

Key words: global localization, global feature points, matching, mobile robots

摘要: 针对传统全局定位方法存在对传感器要求多、计算量大的问题,提出了一种基于全局特征点匹配的移动机器人定位方法。该方法采用普通2D雷达作为传感器,在机器人建立全局地图的过程中同步地提取全局特征点,在全局定位算法中,通过建立局部地图和提取局部地图特征点,实时将局部地图特征点和全局地图特征点进行匹配后求解全局位姿。在两个数据集上的测试,结果优于蒙特卡罗自适应定位(adaptive Monte Carlo localization,AMCL)和Cartographer的全局定位效果,运算速度更快。结果表明,与已有的方法相比,该全局定位方法能够更快地完成全局定位和有效减少计算资源的消耗。

关键词: 全局定位, 全局特征点, 匹配, 移动机器人