Multisensors Fusion SLAM-Aided Forest Plot Mapping With Backpack Dual-LiDAR System

The extraction of forest vertical structural parameters plays a crucial role in forest inventory. In recent years, light detection and ranging (LiDAR) has been widely applied in forest inventories due to its powerful 3-D reconstruction capabilities. The backpack laser scanning (BLS) is a lightweight LiDAR platform that significantly enhances the efficiency and accuracy of forest inventory. To address the issues of Global Navigation Satellite System (GNSS) signal occlusion and LiDAR scanning blind areas under the canopy, a multisensors fusion Simultaneous Localization and Mapping (SLAM) algorithm has been proposed, and a BLS device has been set up. The proposed SLAM algorithm fuses both horizontal and vertical LiDAR data by extracting the planar surface models. In addition, the proposed similar stem features are added to the feature point extraction in forest mapping. The accuracy of the results is validated through standing tree position, diameter at breast height (DBH) and tree height. When compared to other classic SLAM methods, the proposed method achieves 100% accuracy in standing tree extraction, reduces the error in DBH extraction by 85.56% (with an error of 2.05 cm), and decreases the error in tree height extraction by 83.44% (with an error of 0.79 m). The results show that the problem of poor GNSS under the canopy can be effectively addressed by the proposed SLAM algorithm in the study. Furthermore, multisensor data fusion and stem feature addition can provide more complete data support and more robust matching constraints, ultimately resulting in more accurate point cloud mapping.