A Guide to Drone Survey Data


On this page, we talk you through and show you exactly what you need to utilise UAV's in to your surveying workflow. From how to capture the data, through to the final product, we aim to make it as simple as possible.

Contents
Animated markings image
1 +
× DJI L1

The latest LiDAR scanner from DJI

See more
2 +
× DJI Matrice M300

DJI's latest and most advanced enterprise UAV

Go to bundle
3 +
× DJI TB60 Extra batteries

Go to Item
× DJI L1

The latest LiDAR scanner from DJI

See more
× DJI Matrice M300

DJI's latest and most advanced enterprise UAV

Go to bundle
× DJI TB60 Extra batteries

Go to Item

Why should I use a drone?


Widely spread images and videos show realistic 3D models of scenes such as buildings, construction sites, and terrain that have been generated by drones. These data are also used to generate other useful products for users such as contour lines, digital elevation models (DEMs) and volumetric measurement reports. But how are these data generated, and what data do drones actually capture?

There are two main types of sensors mounted on drones: RGB sensors (cameras) and LiDAR sensors. 

Animated markings image
1 +
× Zenmuse L1

The latest LiDAR sensor from DJI

More
2 +
× Zenmuse P1

The latest LiDAR sensor from DJI

Mehr
× Zenmuse L1

The latest LiDAR sensor from DJI

More
× Zenmuse P1

The latest LiDAR sensor from DJI

Mehr

LiDAR

LiDAR sensors emit thousands of laser pulses per second, which are reflected back from the target surface to the sensor. Based on the laser pulse’s time of return, the distance, and position, can be calculated. This is repeated for each pulse, collecting data for millions of points, such as position and intensty. This data is processed in specific software with positional data from the drone to generate an accurate georeferenced point cloud, similar to the point cloud from photogrammetry, which also usually contains millions of points. Some lidar sensors also come equipped with an RGB camera allowing the lidar points to be colourised in real colour.

RGB

RGB sensors are able to capture a scene in a way most similar to human vision, put simply, they capture regular photos. To generate a 3D product, the photos, which have a high degree of overlap with each other, are processed with photogrammetry software that sees common pixels in images, and with a bit of magic, generates 3D points. Millions of these points are usually generated from projects with a few hundred photos. From this point cloud, a surface can be additionally created based on the positions of the points, creating a 3D model.

What can I do with drone survey data?


So, you have a point cloud containing millions of points, or a 3D surface model, but how can these data be used? What useful products can you get from these?

A list of many of the possible deliverables and use cases is shown below:

  • Creation of point clouds, 3D models and orthophotos
  • Creation of digital earth and surface models
  • CAD drawings
  • Data/user portals for project management
  • Full project support of the customer (consulting, equipment, training and project implementation)
  • Calculations of length, area and volume
  • Evaluation of the accuracy of the results by independent measurements
  • Documentation - position of assets
  • Reports on power line-vegetation hazards, and sag analysis
  • Topographic survey
  • As-built survey
  • Facade scanning
  • Asset inspection and surveying
  • Calculation of earthworks and stockpile volumes
  • Monitoring, control and deformation measurements
  • Inspection of wind turbines
  • Pipeline and bridge documentation
  • Thermal inspection
  • Power mast inspection

The following section will go more into more depth for some of the specific data that are required for many of the above listed use cases.


A point cloud of a warehouse captured from a drone.

DEM (Digital elevation model)

A DEM is a common product created from drone data. It is simply a file comprised of pixels, like a photo, but each pixel contains elevation information, representing the surface. A DEM actually shows the shape of the terrain without the surface objects e.g. buildings, whereas a DSM (Digital Surface Model) would include all surface features. These values are often colourised according to the level of elevation providing a product like shown below. Alternatively, the pixels can be given real-colour to appear like a photograph. A DEM gives a good impression of the shape of the land. 

What exactly you may use a DEM for, is dependent on the specific project and desired outputs. However, it is a very common data format in GIS tasks, used either as an end product, or a preliminary one in order to achieve the desired product. For example, they can be used for modelling water flow and where water is likely to pool, line of sight analysis (what is visible from a certain point), or visualising archeological areas of interest. They are particularly useful for large-scale surveys where a point cloud would require an enormous amount of storage and high computational power to display. On the other hand, a DEM is relatively very small in data size, and is much easier to display. 

The image to the left is a DEM of a construction site. Yellow shows higher elevation, purple shows lower elevation.


Contours

Contours are lines that trace along a line of equal elevation, e.g. 10 m, 15 m, 20 m and so on. They’re useful because they make clear the shape and slope of the terrain, which is perhaps not as clear in a DEM or point cloud. This helps, for example, if you are planning on building on the land, and you want to see if the land is suitable regarding elevation and slope, or to see the extent of cut and fill works required. Additional products such as a slope raster, or hillshade raster can also be made based upon the original DEM, as shown below.

The image to the left depicts contour lines generated from a point cloud, overlaid on a DEM. They can also be generated from a DEM. The height of each contour line is stored and can be displayed if desired.

Pictured above is a slope raster based on a DEM, lighter colours show areas with the higher slopes, e.g. walls and trees.

Pictured above is a hillside raster based on a DEM. This simulates shade from a light source, giving another perspective on the shape of the terrain.

Volume measurement

A very useful calculation that can be done with drone captured data is volume measurements. This is useful for many applications. For example, if you plan on developing a site and would like to know the optimal elevation to build on with regard to cut and fill, a volume measurement tool can show you at which elevation the area of interest would have an equal cut and fill, increasing the efficiency of the earthworks operation. Additionally if you plan to dig or raise the land to a particular elevation, the estimates of cut and fill are provided.

The image to the left is measuring the cut off and fill of a possible development at a particular elevation.


Stockpile measurement

Another key use would be measuring the volumes of stockpiles. Traditionally, these are very difficult to measure to a high level of accuracy. However, with precise measurements from drone sensors, such as the L1 and P1, tools can be applied to give a quick and efficient volume measurement of the stockpile. This is a very efficient method that could be widely used in mining and construction.


Classification

Point cloud classification allows users to classify what each point in the point cloud represents. Common examples are ground, vegetation, buildings, power lines, roofs etc. Points of a particular classification can be extracted to optimise post-processing operations. For example vegetation could be removed from the point cloud to allow for easier volume measurements, or ground points can be extracted to generate a bare-earth DEM from the original DSM of the area.

Further post-processing techniques can also be applied once points are classified. One example: powerline inspection, is described below. Another example is forestry analysis. Individual trees can be segmented from the point cloud, and various statistical analyses applied to them. This would be useful for many applications such as forest managers (for timber), forest fire management, conservation, forest health monitoring, carbon sink analysis and more.



Powerline danger points

Also founded upon point cloud classification, points representing power line towers and cables can be classified, and then vectorised to allow for danger point analysis. This compares the positions of classified vegetation points to the power cables, and then, based on a user defined threshold distance, highlights which trees need to be trimmed to avoid possible contact, which could otherwise lead to power outages and forest fires if left unchecked. Powerline sag analysis can be performed easily and efficiently with drone LiDAR data, such as the one from the L1. Additional accessories, such as the in-built stereoscopic cameras and the CSM radar, prevent collisions and provide visual distance information to safely perform drone flights in proximity of dangerous assets.

How can this be achieved?


Now that you know what can be achieved with drone-captured data, you’re probably wondering how it can actually be done.

The equipment


To capture the data there are a number of solutions available. DJI have multiple drone platforms and payloads capable of capturing data for these products. The Zenmuse L1 and P1 are the current flagship payloads from DJI, which are compatible with DJI’s flagship enterprise drone: the Matrice 300 RTK (M300 RTK). The Zenmuse P1’s 45MP 1” CMOS sensor is DJI’s highest resolution RGB sensor, suitable for highly detailed photogrammetry surveys. DJI report the P1 to be capable of achieving an absolute accuracy of 3cm in the horizontal axis and 5cm in the vertical axis. This was measured from results generated from photos with a 3 cm Ground Sampling Distance (GSD). However, test flights conducted by Epotronic found that, when flying at 50m altitude (AGL), which provides a GSD of 0.63 cm, accuracies of 1-2 cm were achieved in the horizontal and vertical axes in DJI Terra.

The Zenmuse L1 is DJI’s first lidar sensor, which is officially capable of achieving a horizontal and vertical absolute accuracy of 10 cm and 5 cm, respectively, is also an attractive solution. However, in tests conducted by Epotronic we measured vertical absolute accuracy as 2.1 cm from a 50 m AGL flight, and 5.1 cm from a 100 m AGL flight, along with vertical precision of +/- 3 to 5 cm, and +/- 7 to 11 cm, respectively. Despite the lower accuracy of the L1 compared to the P1, the LiDAR sensor holds advantages over the RGB solution, such as being able to work in low- or no-light, lower processing requirements and data storage, and the ability to capture detail behind some objects, e.g. glass or trees.

The Phantom 4 RTK (P4RTK) utilises an RGB sensor like the P1, although is of lower resolution. This means it must be flown closer to capture comparable resolution, and is therefore less time efficient. However, it can still capture sufficient detail and at high accuracy thanks to its RTK module, achieving horizontal and vertical absolute accuracy of approximately 2 cm.


Animated markings image
1 +
× DJI Zenmuse L1

More
2 +
× DJI Phantom 4 RTK

More
3 +
× DJI Zenmuse P1

More
× DJI Zenmuse L1

More
× DJI Phantom 4 RTK

More
× DJI Zenmuse P1

More

The processing


The initial processing of both the photos from RGB sensors, and raw lidar data into workable point clouds can be achieved in DJI Terra. As this software is from the world leader in drones and drone sensors, the workflow when using DJI drones and sensors is seamless. DJI Terra has a very simple user interface, allowing the user to import the photo or lidar data, e.g. from the Zenmuse P1, Phantom 4 RTK, or Zenmuse L1, change a few settings such as output coordinate system, resolution, and marking ground control points in the case of photos, and then the software can generate the point cloud. An optional model can also be created for photogrammetry point clouds. The point cloud can be viewed, measured (measurements of points, lengths, areas and volumes, which can all be saved and exported) and annotated in DJI Terra. The point cloud and 3D model can be finally exported in a variety of common file formats such as LAS. For details on the exact processing steps required, please refer to Epotronic’s tutorial videos found at the following links below:


3D model generated in DJI Terra

To generate the products described, a variety of post-processing software can be used. For all software, the exported products from DJI Terra can be directly imported.

LiDAR360, from GreenValley, can generate many of the products described. Creating a DEM, DSM (the difference being that a DEM represents the bare-earth, and a DSM represents everything on the surface), and contours with a couple of clicks. A variety of different settings can be applied, to suit the users requirements. LiDAR360 is also capable of classifying points, being able to automatically generate ground points, and apply machine-learning algorithms to classify all points in a point cloud based off of a small sample that the user manually classifies. This allows tools such as tree segmentation, and forest statistical analyses to be carried out. Volume measurements can also be made using a variety of methods, with cut and fill values displayed, which can be exported.

Another programme from GreenValley, LiPowerline, is also able to highlight the danger points in vegetation as previously described. The software is also very user friendly, only requiring a few steps before reports can be exported, specifically designed to guide the team responsible for the vegetation trimming. Other features, such as sag analysis, weather simulation, vegetation growth and tree fall prediction models are also available, giving further warning of future risks.


Powerline danger point analysis. Vegetation that are two close are highlighted in red.

TerraSolid is another programme capable of performing over 1000 post-processing tools to point clouds, such as calibration, registration (stitching together point clouds), powerline analysis, forestry analysis, road and railway analysis, and many more. An image of automatic object detection is shown below. This is also available for free to customers of the Zenmuse L1 until the end of 2022.

The point clouds can also be imported into AutoCAD, where 2D CAD drawings can be produced, which is still a common desirable output even in this 3D world. The data can also be used to create BIMs and Digital Twins by importing into programmes such as AutoCAD Revit. For more on that see our other article: “How to Use Drones for BIM Workflows and Producing Digital Twins”.

Data from drones can be used in a magnitude of different ways. Whatever you are looking to achieve within centimeter-level accuracy, it is easily possible. Survey-level accuracy can be achieved by professional surveyors or should be verified with traditional equipment or by professionals. Common industry formats can easily be exported with a wide range of post-processing software available, such as DJI Terra, to be imported in various software to process georeferenced 3d models or point clouds.


Conclusion


The capabilities of drones can revolutionise the workflow of many industry professionals such as surveyors, architects, and project managers. The flexibility and wide array of data generated by drones, allows it to seamlessly integrate into current workflows. Recent releases from top manufacturers such as DJI, allow for the most efficient workflow yet, from planning and executing to producing deliverables, all possible from just one software programme, or just two for a wide range of post-processing tools included as well. Investing in drone systems will launch companies and industries as a whole into the digital age, with the potential to significantly improve the efficiency of current project budgets and timelines.

To browse the solutions described here and more, please visit our online store at store.epotronic.com, where you can also find lots of relevant information in our blog.

If you would like to get in touch with any questions or queries, please don’t hesitate to get in touch at info@epotronic.com.


Angebot sofort erstellen, 
herunterladen und teilen, 
freigeben lassen und bestellen. 
So gehts:
Artikel in den Warenkorb legen. Unter dem Button Kasse auf "PDF Angebot erstellen" klicken.