农机化研究2025，Vol.47Issue(9)：15-21,7.DOI:10.13427/j.issn.1003-188X.2025.09.003
基于因子图融合地图的果园机器人定位方法
Localization Method for Orchard Robot Based on Factor Graph Fusion Map
摘要
Abstract
Aiming at the complex orchard environment where GPS signals are easy to be blocked and fruit trees are plan-ted with a high degree of similarity,which leads to difficulties in initializing the global position of the orchard robot,in-consistency between the localized position and the actual position,and low localization accuracy,proposed a global locali-zation method for orchard robots based on the fusion of factor graph with 3D point cloud prior map.Firstly,based on the 3D point cloud prior map of the orchard,a combination of GPS-based segmentation of the local map and NDT matching was proposed to estimate the global initial position of the robot in the point cloud map;Secondly,the laser odometry was calculated as the priori position,and the NDT algorithm was used to calculate the global position of the laser point cloud matched with the map;Lastly,a factor graph was constructed to fusing the laser odometry factor,IMU preintegration fac-tor and map feedback factor,the factor graph optimization was executed to correct the cumulative drift of laser odometry and improve the accuracy of global positioning.The experimental results showed that in the complex orchard scene exper-iments,the lateral root mean square error of the global localization trajectory of this method was 0.17 m with a standard deviation of 0.09 m,and the longitudinal root mean square error was 0.12 m with a standard deviation error of 0.08 m,which met the requirements of the robot to perform autonomous high-precision global localization for the orchard task.关键词
果园机器人/点云地图/因子图/全局定位Key words
orchard robot/point cloud map/factor graph/global localization分类
农业工程引用本文复制引用
何创新,冯威,李云辉,欧芳,李楠,苗中华,韩增德..基于因子图融合地图的果园机器人定位方法[J].农机化研究,2025,47(9):15-21,7.基金项目
上海市科技兴农项目(2020-02-08-00-09-F01466) （2020-02-08-00-09-F01466）
上海市科委科技计划项目(22511101800) （22511101800）
国家重点研发计划项目(2022YFD2002400) （2022YFD2002400）
