Absolute 3D Accuracy Assessment of UAS LIDAR Surveying International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

Surveying an area with small, unoccupied aerial systems (UAS) equipped with a lidar mapping payload—absent permanent, stable, geometrical reference surfaces—demands accurate, repeatable data collection procedures. While relative error within a single UAS lidar dataset may reveal itself in strip misalignment, absolute error (particularly horizontal error) can prove more difficult to detect, casting doubt upon the quality of both individual surveys and time change analyses of multiple surveys of the area. To gain insight on the UAS lidar error budget, this study presents an analysis of multiple UAS lidar surveys over a set of accurately surveyed geometric checkpoints. Each flight’s trajectory was processed multiple times using multiple static GNSS base observations, both autonomous and set over surveyed monuments, at varying distances from the study site. Custom algorithms were used to mensurate the geometric targets detected in each UAS lidar survey's point cloud, allowing for precise comparison of both absolute horizontal and vertical accuracy of each survey against the rigorous ground survey. The results of the analysis suggest that high horizontal accuracy can be achieved under a variety of conditions, whereas vertical accuracy is sensitive to the quality of ground control. and a discussion of the results explores the ultimate goal of isolating and understanding the sources and magnitudes of error in the UAS lidar error budget. Grant no. NA18NOS4000198

针对配备激光雷达（lidar）测绘载荷的小型无人航空系统（UAS）开展区域测绘时，若缺乏永久性稳定几何参考面，则需采用精准且可重复的数据采集流程。尽管单份UAS激光雷达数据集中的相对误差可能表现为条带错位，但绝对误差（尤其是水平误差）往往更难被察觉，这会让人对单次测绘成果以及该区域多次测绘的时间变化分析的质量产生质疑。为深入探究UAS激光雷达的误差预算，本研究针对多组经精准测绘的几何检查点开展了多次UAS激光雷达测绘分析。本次研究中，每架次的飞行轨迹均采用多组静态全球导航卫星系统（GNSS）基准观测数据进行多次解算，这些观测数据既包括自主式基准站，也包括设置在已测绘纪念碑上的基准站，且与研究区域的距离各不相同。研究采用定制算法对每份UAS激光雷达点云中检测到的几何目标进行量测，从而能够将每份测绘成果的绝对水平与垂直精度，与严格的地面测绘结果进行精准比对。分析结果表明，在多种条件下均可实现较高的水平精度，而垂直精度则对地面控制点的质量较为敏感。此外，本次研究还对结果展开了讨论，旨在达成分离并明晰UAS激光雷达误差预算中各类误差的来源与量级这一终极目标。资助编号：NA18NOS4000198