Guest Column: Drone Mapping — The Full Workflow
When starting a drone mapping program, many surveyors begin by deciding which drone to purchase, and only after that do they start thinking about how they will produce their final survey deliverable using the drone. This approach often winds up costing months of valuable time as they try to figure out how to produce the deliverables they need.
Instead, it is far more efficient, effective and profitable to start from the end and identify the workflow for creating the final deliverable, then implement the drone program that enables that outcome. Aerotas has used this approach with dozens of surveyors around the United States.
The first step in the drone-to-map workflow is to ensure that the correct data is collected in the field. Then, the field data is processed to produce an orthophoto and digital elevation model. Finally, a surveyor uses that 3D model to create an AutoCAD (or similar) surface and line-work and their final survey.
Here are three steps we’ve identified in the process:
Collect the Right Data In The Field
For field crews to collect the right data, they need to understand best practice for setting ground control, have autopilot software that is optimized and set up for survey-grade mapping, and – critically – be trained on operating procedures that ensure every detail is met in the field.
Setting drone ground control is fundamentally the same as working with an aerial photogrammetrist. Survey targets are set and surveyed around the project site. While more ground control will mean higher accuracy, we’ve found in testing that the ideal is five targets per flight area – one near each corner and one near the middle – with additional targets as needed in significant outcroppings and high or low spots.
Once ground control is established, the drone autopilot is set to slightly overshoot the control on every side, capturing two photo lines beyond each control point. Setting the autopilot involves using a Google Earth-type map interface to simply trace the project area, then setting flight altitude. With the right drone and the right autopilot, the entire process of preparing and launching the drone should take less than five minutes and require little more than pressing one button.
To maximize time saved and ensure reliable data collection, field crews need to be well trained on drone mapping operating procedures. Though the overall steps are logical, there are dozens of specific procedures that ensure safety and make sure the right data is collected.
Stitch Photos Into Orthophoto/Digital Elevation Model
The next step is to process the raw geotagged photos into the orthophoto and digital elevation model (DEM). There are several software solutions for this process, as well as companies that provide processing as a service. While this process is highly technical, it follows the same underlying logic as conventional photogrammetry. Namely, overlapping photos are stitched together based on ground points shared across overlapping photos.
One of the primary differences between conventional and drone photogrammetry is how calibration is done. Conventionally, large and highly calibrated cameras are used so that, with only two images and few tie points, accurate elevation data can be extracted. Drones, however, use smaller and uncalibrated cameras. This is compensated for by taking many photos with very high overlap. With this approach, any single point on the ground is captured in nine to 16 photos. The software then uses image recognition technology to identify thousands of “tie points” shared across photos. Then the software uses those points to stitch the photos together and extract elevation data due to the slightly different angles of each photo.
The drone image stitching software first creates a point cloud, then processes it into an orthophoto and DEM. The DEM is the elevation information that the orthophoto is draped over, creating a 3D digital surface model. After incorporating the ground control points, any point on this digital surface matches (within accuracy tolerances) the northing, easting and elevation of that point in the real world.
Extract Survey Surface And Line Work
This last step is where we see many survey firms stumble when trying to build their own drone program. Though 3D modeling programs like Civil3D are technically able to handle any 3D file, they are not designed for the large and detailed drone-generated surface models. That makes them slow and cumbersome. This is where post-processing solutions come in.
Post-processing software allows a surveyor to effectively conduct a survey as if they were in the field, but do so directly on the digital surface model. The surveyor takes survey points by simply clicking on the digital surface, and each point is recorded by the software as a set of coordinates. Survey points can be collected in layers that match the surveyor’s Civil3D (or Carlson, or other) layering conventions. In this way, when these points are exported from the post-processing software, they open in Civil3D as if from a standard GPS rover or total station.
Focus on the Workflow
With a complete workflow like this, surveyors benefit from huge savings in both time and money, often reducing field time by as much as 80 percent on topographic mapping projects. While a proficient surveyor may be able to capture 60 survey points in an hour, post-processing software can enable capturing 60 points in seconds.
With benefits like these, it is no surprise that surveyors everywhere are rapidly adopting drone mapping technology. When starting a drone mapping program, remember to focus first on identifying the workflow that will produce the needed deliverable as efficiently as possible.