Orchard Navigation Method Based on RS-SC Loop Frame Search Method and SLAM Technology
Ning Xu, Qingshan Meng, Fengping Liu, Zhihe Li, Guangming Wang, Na Guo, Wenxuan Wu · IEEE Access · 2025
With the rapid development of agricultural intelligence, the application of intelligent agricultural robots in orchard management has been widely concerned. However, the complex environment in apple orchards, such as the occlusion of branches and leaves, dynamic changes and differences in lighting conditions, poses great challenges to the positioning accuracy and map construction ability of synchronous positioning and map construction technology. In order to solve the problem that existing synchronous localization and map construction techniques are not accurate enough in this kind of environment, a loopback detection algorithm based on neighborhood search and scanning context fusion is proposed. The method realizes rough loop frame search through neighborhood search to improve speed, and global precise search combined with scanning context to improve accuracy. In the loop frame matching, an optimization algorithm combining normal distribution transformation and iterative nearest point is used to reduce the cumulative error significantly. According to the environmental characteristics of apple orchard, the research preprocessed the point cloud data collected by LiDAR, and combined with the LiDAR odometer method based on line and plane feature point matching, the pose estimation and map construction were carried out. The experimental results show that compared to the independent use of Radius Search and Scan Context, the proposed algorithm increases the number of loopback frames by 50.19% and 7.23%, respectively, and reduces the root mean square error by 28.31% and 45.60%, respectively. The research results significantly improve the positioning and mapping performance of synchronous positioning and map construction technology in the complex environment of apple orchards, and provide technical support for the application of intelligent agricultural robots in orchard management.