Maintenance Products

Commissioning, interpreting the right drone data

Unmanned aerial vehicles (UAVs) – also known as remotely piloted aircraft (RPA), but more commonly known as drones – have certainly had a big impact on surveying in a quarrying environment.

At the CMIC 2016 conference in Melbourne last October, almost everyone I spoke to wanted to know about drone technology.

UAVs are great tools in some instances. Drones can also be used for survey work, as well as manned aircraft, land surveyors and many other types of technology. The most appropriate technology or combination of technologies for any job depends on the budget, timeframe to complete, accuracy needed, safety, legislative requirements and access.

Drones are great! For example, they can be lower in cost compared with manned aircraft and they allow for surveying inaccessible or unsafe areas and may have a quick mobilisation. However, the data produced from drones is not always suitable for use in all instances.

Drones may also appear inexpensive, but there are hidden costs that need to be taken into account when determining if they are the best technology for a particular application. This article discusses the data deliverables from drones, accuracies and hidden costs.

Data deliverables

The basic data obtained from drones is an ortho-photograph – a photo that has all distortions removed and allows distances and angles to be measured directly from the photo.

With some more processing of the data, another deliverable is a digital surface model (DSM). A DSM represents the Earth’s surface and includes all objects on it, such as the tops of trees, buildings, grass, fixed and mobile plant and people.

{{image2-a:c-w:635}}

A DSM cannot be used for engineering design and volume calculations because it doesn’t provide an accurate reflection
of the Earth’s contours.

After a lot of semi-automatic and manual processing, a digital terrain model (DTM) may be generated. In contrast to a DSM, the DTM represents the bare ground surface without any objects such as plants and buildings. Figure 1 shows the difference between a DSM and a DTM. A DTM may be used for engineering design and volume calculations, but the file sizes are enormous. For a medium-sized quarry the file size can be greater than 500MB.

To get UAV data into geology and quarry planning software such as Surpac, further semi-automatic and manual processing is generally required.

This further processing involves extracting lines and strings from the DTM. The lines and strings model ground relief features such as tops and toes of batters, benches, tracks, roads, water bodies and surface drainage lines. Detail features such as fences and buildings are also extracted from the DTM. These extracted line strings generate a file size of about 10MB and may be imported easily into different software applications.

Table 1 shows the different UAV deliverables.

Summarising, drone data can be great for engineering and volume calculations if many hours are spent cleaning up and processing the data before use. Figure 2 shows drone data supplied by a client unhappy with drone survey results because of insufficient processing.

Figure 3 shows the same area after further processing to reflect real world conditions. Note the smooth contours.

{{image3-a:c-w:635}}

Accuracy

A lot of people ask about the accuracy of drone data. The accuracy depends on a lot of factors:

  • The inherent camera accuracy. Metric cameras have all known distortions calibrated. Most drone cameras do not use metric cameras, therefore all camera distortions are propagated through the ortho-photo and models.
  • The stability of the flight. As most drones do not use an inertial measurement unit or a gyroscope, bumpy flights mean the orientation of each image is not known but is derived through software. All errors are therefore distributed through the model.
  • The quality of the on-board GPS data. Some units use a process called RTK initialisation to determine the camera’s position and height. In perfect conditions, the accuracy of this RTK solution is about +/-3cm. The majority of drones just use a consumer GPS with accuracies of several metres.
  • The number and quality of ground control points. More ground control points mean better accuracy, even with RTK drones. More ground control points allow a better estimate of the resultant accuracy.

Each of these elements contributes to the total final positional accuracy of the data.

Resolution and number of pixels do not indicate accuracy. A common mistake is to use ground sampling distance (GSD) as a proxy for accuracy. GSD is the distance between two consecutive pixel centres measured on the ground. For example, a GSD of 5cm means one pixel in the image represents linearly 5cm on the ground. This does not mean that the pixels are accurate to 5cm. The only way to confirm accuracy is to compare check points derived from the drone data to independently surveyed ground check points that have not been used as ground control points.

With good conditions, sufficient overlap of photos and proper ground control, horizontal accuracies of about two x GSD and vertical accuracies of about three x GSD may be achieved. So, if you have a GSD of 5cm, possible best accuracies are +/-10cm horizontally and +/-15cm vertically if proper processing and error elimination is completed. This can only be checked by comparing to ground survey check points.

Hidden costs

Sometimes it may appear that the cost of a drone survey is a lot less than other technologies, but there are other costs to a business that should be determined to evaluate the true cost, eg:

  • As drone pilots have to be on the site they are working in, the drone pilot needs to be inducted by quarry staff.
  • Many sites require quarry personnel to escort solo contractors around the site.
  • Ground survey requirements must ensure accurate models need to be allowed for.
  • If you don’t specify a bare earth DTM as a deliverable, quarry technical staff have to spend hours cleaning up drone data to remove structures, plant, vehicles and trees from supplied drone data.
  • As the drone data file sizes are enormous, the data is extremely slow to transfer and load into software programs, and sometimes causes software programs to operate more slowly. The processing of this extra data currently remains laborious and inefficient.
  • The massive file sizes have negative implications in terms of data storage within an organisation and data back-up.
  • Possible lost productivity, as low flying drones may distract site workers.

{{image4-a:c-w:635}}

In conclusion, drones are great for certain applications. We love them and we use them!

The next time, however, you ask for a drone to survey your site, you should be aware of what data deliverables you need, the accuracy required and some of the hidden costs. For any project, the most appropriate technology or combination of technologies should be chosen. 

Erik Birzulis is the managing director of Landair Surveys.

Leave a Reply

Send this to a friend