The use of drones in surveying has developed rapidly in recent years. More and more often, professional providers are using drones specifically for surveying and mapping work. Thanks to the rapid operational readiness of the technology in conjunction with a high degree of automation, data acquisition has now become child's play, even for non-professionals. Nevertheless, you should bear in mind the most important boundary conditions, because not all of us are surveyors.
Where are drones used?
Basically, there is no limit to the possible fields of application. New areas are being discovered by users themselves every day. However, the mapping of linear transport systems (railway tracks, roads) and construction progress documentation have become established. Volume calculations in open-cast mining and landfill sites are also among the most important areas of application. Drones are also frequently used to map agricultural areas or to document damage in forestry. Furthermore, facades can be documented and measured with a very high resolution. Drones are also increasingly being used in monument preservation, building inspections and BIM.
A 3D model is created from individual images
A 3D model of an object can be created using a special method of triangulation in photogrammetry based on many aerial photographs of an object taken from different perspectives.
The size of the object is initially irrelevant. Starting with small buildings or facades, larger areas covering several hundred hectares can be combined. The images taken must overlap in certain areas. Easy-to-use flight planning software specifically for drone surveying is very helpful here. The images are processed in the computer and common points are merged into a large model. This 3D model can be compared and linked with the data from a laser scanner and can then be further processed, depending on the application.
Control points or RTK drone?
Before the actual flight, the control points are distributed as homogeneously as possible over the area to be surveyed and measured using a GPS rover or total station. If the 3D model is to be transformed into a local or global coordinate system, ground control points (also known as aerial markers) are absolutely essential. This is independent of whether the drone has an RTK function or not. RTK (real time kinematic) naturally has a clear advantage. On the one hand, significantly fewer control points are required, and on the other hand, the time-consuming calculation process is accelerated afterwards. For larger projects, this is crucial for economic efficiency. If you dispense with control points altogether, you still get a good 3D model, but subsequent georeferencing is often error-prone and inaccurate.
Resolution vs. accuracy
There is a wise saying in surveying technology: "Not as accurate as possible, but as accurate as necessary!" The parameters "resolution" and "accuracy" play a major role here.
- Resolution describes the level of detail of the model, expressed in centimetres per pixel. The focal length, distance or altitude as well as the number of pixels (megapixels) of the camera sensor have an influence on this value.
- The accuracy, on the other hand, is a geodetic quantity and can never be better than the resolution. Dealing with the question "How accurate is my measurement?" is very important for a good result.
Example mapping: The result of a flight from a height of 100 metres with a 20 megapixel camera drone has a resolution of 2 - 3 cm/pixel in the best case. If control points are used correctly, the model has an accuracy of 5 - 15 cm in position and height.
Example facade: When measuring a distance to an object (5 - 10 m) with a drone (20 MP camera), a resolution of less than 5 mm/pixel is obtained. Small cracks and damage become visible. In many cases, accuracy is not even required.
Procedure for surveying from the air
1. Flight planning
All necessary authorisations for the flight must be obtained. Precise flight planning is carried out using a polygon for the area to be flown over, with suitable software or an app. As a further preparatory measure, the weather forecast should not be ignored.
2. The control point - laying out and measuring
Depending on the size of the object, more or fewer control points are laid out and measured. For frequent use, we recommend using control points with aluminium composite material; alternatively, sheets of hard foam PVC can also be used. Experience has shown that the control points should be positioned at the outer edges and within the area to be measured. The distance between the points should be a maximum of 100 metres. Smaller control points can be glued to facades or buildings and measured with a total station.
3. The flight
For the actual flight, the automatic GPS flight mode is usually used to take regular images with the desired overlaps. The sharpness of the images and the recognisability of the ground control points should then be checked.
Example 1: Recording a square area of 10 hectares from a height of 100 metres takes approx. 20 minutes with a DJI Phantom 4 RTK. Shooting facades or buildings requires the use of automatic release functions, as flight planning on vertical objects is almost impossible.
Example 2: Shooting a façade with 4 storeys and a length of 30 metres takes approx. 15 minutes with an interval triggering of 2 seconds. Please note that good flying experience is required, especially in city centre areas. We recommend drone training with a guided practical session. The drone flight should always be carried out in compliance with the legal requirements. You should also always keep an eye on the weather. Distance sensors can no longer react correctly, especially when flying close to buildings, as reflections can occur in the windows.
4. Processing the images with photogrammetry
The images are loaded into photogrammetry software together with the GPS position data (Exif data) and merged into a 3D model using graphical calculations. The control points are also added to improve the absolute accuracy or to transform the data into a local coordinate system. The resulting point cloud can now be used as required, for example in GIS or CAD programmes.
5. Exporting the results
It is often sufficient for the user to generate a so-called orthophoto of the terrain in order to obtain an up-to-date and detailed overview of a terrain or construction site. The 3D model can now also be used to determine volumes in open-cast mines, spoil tips or landfill sites.
Requirements for the computer
The PC has a lot to do to generate such a point cloud. Both the CPU and the graphics card are heavily utilised during the calculation phase. It is often helpful to start the calculations in the evening. Weekends or public holidays are also popular "calculation days". If you don't have a suitable computer and don't want to buy one, you should use cloud services such as PIX4D or Dronedeploy. The images are simply uploaded there and the calculation takes place on a server. However, there are often restrictions with regard to the parameters. In some cases, errors cannot really be reproduced - you have to take what you get. But the investment, especially for a few projects, is lower, as project-related billing is possible.
If you decide in favour of the desktop PC variant, you should not go below the minimum hardware requirements. As a general rule: the more expensive, the faster and you should not skimp on RAM (working memory). The more images are used for the calculation, the more RAM you need!
Data handling - where to put all those gigabytes?
We've all been there: large amounts of data are often a disaster in everyday life. Sending by e-mail or cloud service providers is usually limited, not to mention the transfer time.
An example: An orthophoto (geoTif) in the highest resolution (1.5 cm/pixel) of an area of 15 hectares has a size of 2.5 gigabytes.
The principle "only as much as necessary" also applies here. For data transfer, you should select exactly which area is to be exported in which resolution. Avoid files that are too large, as this will not satisfy your customers either. Problems can occur when opening large amounts of data.
Limits of photogrammetry
Drone surveying is not an all-purpose weapon and has its limits. Basically, the better the quality of the image, the better the quality of the model. The following parameters play a major role and lead to success:
- Sharp and focussed images
- Homogeneous exposure (avoid long shadows or bright areas)
- Good overlap of the image series (at least 65%)
- Clean camera lens
If it is dark or the objects move between the individual shots, you will not get any meaningful results. Low-contrast surfaces - such as tarmac or white house facades - are also a challenge. No results at all can be expected on water surfaces, as water is constantly moving. Photographing through water is also not easy, as the surface is usually reflective.
In this case, a combination of different surveying methods is often used to achieve a result. The points of a laser scanner can be combined with the point cloud from photogrammetry. Furthermore, there are many intersections with classic surveying (rover and total station)
If the photogrammetry method is used, there are several drones to choose from for aerial photography. In order to find the right one, a few questions must first be answered depending on the intended use.
- How often will I use the system later?
- What object resolution do I need? (1 mm or rather 2 cm/pixel)
- What budget do I have available?
- Do I have suitable flying experience? (Flight without GPS support on facades)
- What size is my emergency vehicle? (sufficient power for recharging the batteries)
We will be happy to answer the most important questions together with you.