From Drone to the Field: Processing and Deploying UAV Imagery
Unmanned Aerial Vehicles (UAVs) are becoming ever more commonplace in the mapping business for collecting cheap aerial imagery on demand. Once an expensive hobby for tinkerers, UAVs have rapidly become a truly disruptive technology in the geospatial industry. The ability to quickly send an inexpensive and nimble platform of sensors into the air to survey your surroundings, and receive realtime feedback, has opened the door to countless opportunities for data collection and analytics.
Like many emerging technologies, UAV advancements have continued to accelerate at a pace far quicker than lawmakers can keep up with. There are very real safety and privacy concerns that need to be addressed, but we are still waiting on reasonable legislation that will permit commercial use of this technology. In the meantime, businesses and tech entrepreneurs are continuing to invest in UAV research and development, so they can maintain a competitive advantage once legislation has been passed.
We see UAV technology as a major opportunity to enhance both the efficiency and the quality of ground-level data collected in the field. In order to keep abreast of this technology, we’ve partnered with some leading players in this space.
UVM Unmanned Aircraft Systems Team
For the past year or so, The Unmanned Aircraft Systems Team at The University of Vermont’s Spatial Analysis Lab has been working on an innovative project to research Unmanned Aerial Systems for Transportation Decision Support. This initiative, which is funded in part by US Department of Transportation, seeks to exploit the data collected by Unmanned Aerial Systems for disaster response and recovery. In short, the DOT is looking to use UAVs as an alternative to current commercial remotely sensed datasets, which are expensive, often out of date, and low resolution. This data is heavily used by state transportation agencies for transportation planning, operations, maintenance, and program development.
The UAS Team at UVM, lead by Jarlath O’Neil-Dunne, has been extensively testing the senseFly eBee fixed-wing UAV platform for collecting and processing aerial mapping data. Their system includes the eBee with associated sensors, mission planning and control software, data processing software, and the computing power and expertise to transform the raw data into 2D orthomosaics (maps) and 3D models.
Here’s a quick video showing the UVM team in action deploying the eBee:
Integration with Fulcrum
As part of their R&D effort to build guidelines for using drones in this context, UVM reached out to the Fulcrum team for a partner with expertise in ground-level field data collection in the commercial sector. The UAS Team realized that the Fulcrum platform could make an excellent mobile delivery system for this type of data, in addition to providing additional vector data collection capabilities.
Fulcrum makes a great platform for ground-truthing UAV derived aerial imagery. After the imagery has been processed and the orthomosaics have been built, you can quickly deploy this imagery to the field for additional ground-level field data collection, including GPS locations, photos, videos, and surveys. The UVM team is also using Fulcrum to record pre-flight checks and other logistical information related to their missions.
Real-World Use Cases
First responder and emergency management organizations would be ideal consumers for UAV-based real-time imagery capture. They also need field tools for conducting things from damage reports and needs assessment for Individual Assistance to Public Assistance applicant verification to get infrastructure back in shape. Collected imagery can be used to verify claims on damage to levees, critical facilities, parks, signs, roads and be used to calculate debris removal required. Part of UVM’s development process is to optimize the stages of the workflow:
- Deploy team after an imagery request
- Put a UAV over the area of interest
- Capture imagery
- Download and process data
- Publish finished products for users
So a logical next step was to add the final stage of push data to field operations personnel for offline reference maps.
The process takes as little as a few hours to go all the way from initial team deployment to a finished, processed imagery product for consumers to use, at the cost of a UAV platform instead of putting an aircraft (and a person) in the sky to capture imagery. The final stage is to process that raw imagery output into something usable in Fulcrum: the MBTiles format. So using both TileMill and MapTiler (we tested both), we generated and uploaded sample images to Fulcrum and downloaded them to some mobile devices. Adding less than another hour, our near real-time aerial photography can be available to disaster response organizations in the field the same day it’s captured.
Here’s the finished product, downloaded and displayed as a base layer on an iPad. This image was taken after a flash flood of the Great Brook east of Montpelier - along the edges of the stream bed, you can see silt deposits, debris piles, and roadway outwash. If imagery like this was available to ground crews, they could more quickly identify areas in need of attention.
Drones As A Service?
In the relatively short time since UVM began testing their workflow, the commercialization of drone processing has gone a step further, with the launch of cloud-based drone data processing services, such as DroneDeploy. DroneDeploy removes the complexity and expenses associated with standing up your own mission planning and data processing software. While the eBee is capable of covering a relatively large area (up to 12 sq. km), the computing power and expertise required to process that much data quickly is not trivial.
DroneDeploy offers an intuitive platform for end-to-end drone data collection. They have a mobile app for mission planning, which allows you to define an area of interest, after which the software connects to your drone (typically a DJI quadcopter), performs all the necessary system checks (battery levels, GPS fix, camera storage, etc.), autonomously flies the mission with the proper photo overlapping, returns to home and uploads the data to their servers for immediate processing. The end result is a readily available web map and exportable GeoTIFF of the orthomosaic, digital surface model, 3D model, and NDVI product.
The processes are still in development, but we’re making some excellent headway in showcasing how readily UAV-based imagery can be deployed on the ground in real-world scenarios for disaster relief operations. Whether you are looking to invest in an in-house system, such as the eBee, or take advantage of cloud services like DroneDeploy, the future has never looked brighter for capturing fresh aerial imagery for immediate decision making.