Team UAV GeoLab from the Warsaw University of Technology, together with drony.net in cooperation with the company Skysnap conducted a series of test flights with a laser scanner DJI Zenmuse L1. Test flights took place in two different areas. The first area was the aggregate landfill by a newly built road near Mława in June 2021. Additionally, on August 9, flights were carried out over the test field for Lidar UAV systems near Warsaw. The nature of the facilities varied; the first area was uncovered, mainly with aggregate dumps and low voltage lines. On the second site were trees, low buildings and elements of the test field.
After obtaining the data in the DJI Terra program, clouds of points were generated. The DJI Terra program is an intuitive program that requires you to select the path to the folder with the input data, indicates the resulting coordinate system and select data processing parameters.
In the next step, after processing the data from the raids, the GeoLab UAV team conducted a series of experiments and analyses based on the obtained data and geodetic field measurements. Examples of analyzes are point cloud density depending on flight altitude, speed, frequency and scanning pattern, point cloud accuracy analysis based on points with known coordinates, height differences between adjacent series (scanning stripes), and standard deviation of point distribution on a flat surface, e.g. on the way. Based on the data, visualizations of aggregate storage sites were also made, which can later be used to measure the volume.
There were low-voltage lines in the aggregate storage area. Due to the high density of data from the L1 scanner, points were also registered on the lines, as shown in the visualization below.
The DJI Zenmuse L1 scanner allows you to record data in two scanning modes: circular (non-repetitive, [a]) and linear mode (repetitive, [b]).
A point cloud has a different point density distribution depending on the scan mode selected. When scanning in circular mode, the point cloud density decreases as you move away from the centre of the scan belt. In linear mode, the cloud density is more constant across the row, with a higher cloud density at the extremities.
In the DJI Terra program, you can select data processing parameters – Optimize point cloud accuracy. When selected, the accuracy of the point cloud is expected to be more significant. A comparison of the height models of the generated two-point clouds was performed: with and without using the Optimize point cloud accuracy function. The differences between the height models for flat areas and heaps are in the range of – / + 10 cm.
Based on the data obtained over the test field in Józefosław near Warsaw, it was possible to analyze, among other things, the passage of the laser beam through trees in the case of three different scanner frequencies, which also differ in the number of registered reflections (registration of a maximum of 1, 2 and 3 beam reflections, respectively). For this purpose, cross-sections through 1 m wide point clouds were made, and all clouds were obtained from the height of 50 m AGL.
The cross-sections show that the canopy of trees is similar regardless of the scanning frequency. On the other hand, the more beam reflections are recorded, the more points are visible in the lower parts of trees and on the ground.
In addition, analyzes were performed based on objects in the test field. The results show that the errors are within 5 cm for most of the objects, and for four planes, these errors were more significant and reached 20 cm.
The results showed that the measurement accuracy with the DJI Zenmuse L1 is acceptable. When comparing the point cloud with the reference measurements, differences of 5-10 cm were obtained. Clouds of points from the L1 laser scanner are characterized by higher noise than other UAV laser scanners available on the market, which are also characterized by a higher price. It is worth noting that the DJI Zenmuse L1 scanner is 2-3 times cheaper than its competitors; hence the noise may be a bit higher. This means a more significant standard deviation of the point distribution on planar points for point clouds in overlapping adjacent stripes also varies from 5 to 10 cm.
Our company, based in the United Kingdom, is made up of a team of selected pilots, all recognized by CAA, who, with extreme professionalism, attention and dedication to their work, constantly strive to offer different services related to the world of RPAS ( Remotely Piloted Aircraft Systems), better known as Drones.