3D LiDAR-Based Semantic SLAM for Intelligent Irrigation Using UAV

Abstract

Ensuring water use and food security is essential due to the growing world population and global warming. Agriculture is the largest consumer of freshwater, and attention has been focused on improving water-use efficiency in irrigated agriculture. We propose 3-D light detection and ranging (LiDAR)-based semantic simultaneous localization and mapping using unmanned aerial vehicles (UAVs) for intelligent irrigation. The proposed system uses the water-absorbing property of LiDAR to define a water point cloud and segment the surface water area based on singular value decomposition. A path is created using random sample consensus as the median point of the divided surface water area. By extracting the width and height information from the surrounding point cloud, the system aids in proactive natural disaster prevention and has potential applications for Big Data. The performance and practical utility of the proposed system were demonstrated via field data using a UAV and 3-D LiDAR. The robustness of the proposed system is verified by experiments in two irrigation environments with different surface water widths and temporal conditions.