Forest in situ observations through a fully automated under-canopy unmanned aerial vehicle
| Autor: | Xinlian Liang, Haiyun Yao, Hanwen Qi, Xiaochen Wang |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2024 |
| Předmět: | |
| Zdroj: | Geo-spatial Information Science, Vol 27, Iss 4, Pp 983-999 (2024) |
| Druh dokumentu: | article |
| ISSN: | 10095020 1993-5153 1009-5020 |
| DOI: | 10.1080/10095020.2024.2322765 |
| Popis: | Close-range sensing has yet to attain the status of being a dependable source for in situ forest information as the conventional field inventory. Each solution has its advantages and disadvantages in terms of accuracy, completeness, and efficiency. For a forest area, Terrestrial Laser Scanning (TLS) has the highest data quality, but is limited to static perspectives and lack the efficiency. Mobile Mapping Systems (MMS) systems gain on the efficiency but compromise the data quality. More recently, under-canopy UAV caught attentions for its potential to leverage the advantages of both TLS and MMS systems. This study demonstrates the feasibility of autonomous forest in situ investigation using an autonomous under-canopy UAV Laser Scanning (ULS) system, and evaluates the performance of such system in deriving key forest and tree attributes through a comparison with other close-range sensing systems such as the TLS and the Personal Laser Scanning (PLS). The under-canopy ULS system uses an onboard LiDAR sensor to aid its self-traverse in an unknown forest environment and to collect point cloud data during its movement inside the forest. Key factors influencing the systems’ overall performance were investigated via various experiments. The point cloud data collected by the under canopy autonomous ULS system deliver similar stem capturing capacity as TLS in single layer forest stands with less undergrowth. The RMSEs of the DBH estimates were 0.81 cm (3.80%), 4.12cm (19.92%), and 5.13cm (22.01%), respectively. The RMSEs of the stem curve estimates were 1.27 cm (5.48%), 3.97 cm (17.63%), and 5.18 cm (22.49%), respectively. The geometric accuracy and the completeness of the point cloud significantly improved when the trajectory was densified. More studies on autonomous route planning in complex unknown forest is required to improve the system mobility, data quality, and the applicability of such systems in future practical forest in situ observations. |
| Databáze: | Directory of Open Access Journals |
| Externí odkaz: |
|