LIDAR for surveying drones: Functionality, advantages and disadvantages

Figure 1: LIDAR mode of operation

Possible objectives of a LIDAR measurement

We now know that a LIDAR can record information about the distance to an object as its composition. But what goals can be pursued with it? In general, the result is a point cloud, i.e. the set of all points that the LIDAR has captured during the mission. Let's now take a look at the different possible applications.

Surface model of a forest floor

Figure 2: LIDAR (top) vs. photogrammetry (bottom) by drone over significant vegetation

Source: Research report [1]

Foliage and branches significantly impair ground visibility over forests, posing a challenge for remote sensing by drone. Although LIDAR can achieve significantly better results than photogrammetry, it cannot (as is often assumed) see through vegetation. LIDAR can only make better use of the existing gaps in the vegetation to still catch a "glimpse" of the ground.

Documentation and investigation of high-voltage lines

Figure 3: LIDAR model with trees and high-voltage power line

Source: True Reality Geospatial Solutions, LLC [2]

Energy companies use airborne LIDAR to obtain geometric information about their infrastructure, such as cable deflection, distances between poles and wearing pylons. The findings on wear and tear and the condition of the corridor are documented in a 3D model and made available worldwide. Safety risks such as increasing vegetation and animal or human influences are quickly detected and quantitatively analysed in CAD. In addition, maintenance and repair work can be better prepared and organised more efficiently.

3D modelling of complex structures

Figure 4: LIDAR model of the shell of a high-rise building

Source: SPAR3D [3]

LIDAR shows its strengths in all construction phases where precision and small details are important. While photogrammetry provides good results for larger distances, LIDAR can bring its advantages to bear in small and angled construction areas. This is the case in shell construction, for example. Once the data has been captured, the 3D model is analysed directly in the CAD system of your choice and used further.

Limits of the LIDAR

Of course, LIDAR is not the solution to all problems. The most important limitations should be mentioned here once again.

LIDAR can NOT see through vegetation!

The advantage of LIDAR is that individual points can be analysed directly and therefore even small gaps in the vegetation can be used to provide information. However, the wavelength used cannot penetrate vegetation.

LIDAR can NOT detect colours!

Although light is emitted and reflected, no statements can be made about the colour of the detected object. You can only estimate the reflectivity of the object. However, in order to generate true-colour point clouds, an RGB camera is usually also installed in a LIDAR. The problem is therefore not usually relevant in practice.

Positioning accuracy of the drone is important! D-GNSS first choice

We now understand that LIDAR can provide good data. But so far we have ignored one crucial factor: the drone! The LIDAR determines the position of each measured point in relation to the position of the drone. It follows that a poorly determined position of the drone inevitably leads to a poor model.

Imagine the following situation: You have been given the task of determining the coordinates of a tower, but you are not allowed to move and are only allowed to use a GPS device and a compass. You know for certain that the tower is 50 metres away. You look at your compass and realise that the tower is exactly north of you. Then read your own location from the GPS device and correct the position by 50 metres to the north. Great!

But what haven't we considered? GNSS receivers can only determine your position to an accuracy of 5 - 10 metres! The consequences are serious: if the GPS has determined your position 7 metres too far to the left, then the coordinates of the tower are also 7 metres wrong. Now imagine you are measuring a row of towers standing exactly next to each other. You do everything correctly, but what happens? The coordinates of the towers vary by 5 - 10 metres in all directions. If you plot the results on a map, the row is no longer a row but a snaking line with wildly distributed towers. And this is exactly the problem that occurs with drones when you fly over an area and record a row of objects. But what is the solution? D-GNSS.

Figure 5: LIDAR relevance of the GNSS position

D-GNSS is a supplementary system to normal GNSS that greatly increases the positioning accuracy of the drone. In this way, 5 - 10 m accuracy can become 1 - 2 cm. And this accuracy is of paramount importance for LIDAR systems! This is because positioning errors are transferred 1:1 to the model, as in the tower example.

So make absolutely sure that your drone either has RTK (= our recommendation) or that you use PPK.

Precise information about camera position and orientation is essential: IMU and trained personnel

The position of the drone is only a basic piece of information for accessing the key information, namely the exact position and orientation of the camera sensor! This is where our example from above comes into play again: If the camera orientation is not inaccurate, then the measured point is in the wrong place. If we now measure a whole series of points, the difference becomes visible in a loss of quality in the model. A LIDAR therefore always has its own IMU , which calculates the exact position and orientation of the camera sensor. The quality of the IMU has a direct impact on the quality of the model. When you buy a LIDAR, a considerable proportion of the price is always due to the development of a sensor system for exact position determination.

How can you guarantee a good quality model? The key is trained employees. Errors can be recognised above all in the post-processing of the images. Some errors are obvious, such as gaps in the model. Other errors - and these are the much more serious ones - occur even though the model appears to be intact. These can be distortions or wave-like phenomena that have their origin in incorrect position or orientation information. A reliable method for validating the model is to use reference points, e.g. by laying out control points whose coordinates and distances are known. If a test measurement in the model shows a serious deviation from the actual distance between two points, errors can be reliably identified. The bad news is that warped models usually cannot be repaired and must be reflown.

Conclusion: Advantages and disadvantages of LIDAR

We summarise that LIDAR can be an effective alternative to photogrammetry. However, the usefulness of an application depends very much on the target to be captured. If you need to record areas that are rich in detail or covered in vegetation, LIDAR is certainly the right choice. The LIDAR also impresses with its resistance to weather , adverse lighting conditions and darkness. You should have a budget of at least €30,000 for this and plan for training costs for your staff.

Do you have questions about the purchase decision? Contact us by e-mail at info@airclip.de or by telephone on +49-351-211 68 690.

FAQ

Question: In which formats is a LIDAR point cloud usually output?

Answer: .las or .laz

References:

[1] https://www.mdpi.com/1999-4907/7/3/62

[2] http://www.liforest.com/?p=14089

[3 ]https://www.spar3d.com/news/uav-uas/hovermap-powerful-slam-drone-autonomy-lidar-mapping/