Drones fly over aircraft accident scenes in UK: mapping wrecks, not causing them
As the UK fights to keep aircraft safe from reckless drone pilots, their air accident investigators use flying robots to document aircraft crash sites.
Stuart Hawkins has used drones and photogrammetry software to document fatal aircraft accident sites for nearly two years now. This senior air accident inspector works at the UK AAIB, a branch of the Department for Transport which investigates an average of 245 aircraft accidents (and serious incidents) annually.
Hawkins spearheaded the drone photogrammetry workflow at the AAIB and was one of the first in the world to use it in the profession. He recently received the ISASI Award of Excellence for Best Seminar Paper for his thorough report on these uses, and we called him up for more details.
You’ve seen a lot of accidents in the last 14 years at the AAIB, including 60 field investigations. What is that like?
It depends on if it’s a fatal accident in a field somewhere. Then you arrive, and you’re first of all met by the police, getting a briefing, trying to work out what might have happened and talking with witnesses. I’m mentally very alert, trying to take in a lot of information quite quickly.
Initially, we’re trying to work out how it hit the ground — did it hit at high or low speed, with wings level, or was it a steep angle of bank? Those sorts of things give initial clues to what might have happened. Sometimes the bodies are still there, which is always very sad to see, and it does make you think of the families. We deal with it as sensitively as we can while we gather evidence from the scene. The drone has been very useful to capture that initial scene before we start disturbing it and potentially cutting up the aircraft.
Your report includes examples of how the AAIB has used drones and photogrammetry software on scene. You’ve been using it for two years now, haven’t you?
We’ve been using drones for two years, and photogrammetry for a year and a half. It’s been very useful to us.
The drone’s usefulness was clear straight away; we used to rely on images taken by a police helicopter, but they didn’t always get the images we wanted, and sometimes we’d have to wait a week or two to get them on a CD. Now with the drone, we can go up, get the aerial shots, take the angles we want, and if we’re missing parts of wreckage we can use the drone to do a live search as well.
Photogrammetry, particularly the 3D model and orthomosaic, has given us another great way of capturing large scenes. Being able to take measurements is very useful too, and saves time on site.
We don’t have to take as many measurements on site, don’t have to take as many GPS waypoints, and you can actually get accuracy with the photogrammetry software that’s better than a handheld GPS. Importing an orthomosaic into Google Earth is great, so then you’ve got the surrounding terrain plus a very detailed image of the accident site.
Was there a moment after you began using this workflow that you first said, “wow,” this really works, or did it come slowly?
My colleagues were positive and open to it, but it wasn’t until we used photogrammetry at the Blackbushe airport accident site that its usefulness became apparent. At Blackbushe, a Phenom 300 business jet had overshot the end of the runway and crashed in a car auction site.
That was the first accident site I used Pix4D at. When we come back after an investigation, we always do a presentation to the whole branch to explain how it went, and what we have found out so far. I presented the 3D visualization of that site, and people went “Wow!” they were quite impressed. It was off the back of that we got full approval to buy the license, as I was running on the trial license. So it was that accident with the trial license that sold itself in terms of a business case, that this would be very useful in the future.
What percentage of the time do you use drones and Pix4Dmapper now?
I guess at half, if not over half, of the fatal accident sites we’re using it. There are some times when it’s a very simple, small accident site, and we don’t get the drone up. And there have also been times when some of my colleagues have just got the drone up for oblique images and not done the grid pattern.
How do you feel when you see drones in the news, as a menace to airplanes?
We talk about it, and usually whenever there’s something in the news it gets circulated by email. We are very interested in it; drones do pose a potential threat to aircraft if not used responsibly. There are clear safety guidelines published by our aviation regulator in the UK called the ‘drone code.’ I know that there’s work ongoing to try and identify what the risk is and potentially some tests will be carried out, impact tests with small drones and aircraft on the ground to see what kind of damage might be caused. We take safety very seriously and a team of four of us created an operations manual for the operation of our drones.
Are there any reasons you think the industry won’t adopt this method?
So some of the obstacles are cost and resource. With cost, you’re investing in the drone and software. The second is related to resources for drone operation. We’re quite fortunate in the UK. We can operate our drone in the same way a recreational operator can, so we don’t have to jump through all the hoops of a commercial operator, although we have gone through some of them voluntarily anyway.
Another thing that helps us is that we have two engineering support guys. One of them will almost always come out to the accident site and they help recover the wreckage. At the beginning, they’re not normally too busy, as we’re not cutting up the wreckage straight away, so they have had the spare capacity to take on the drone flying task. We’re quite lucky to have these two trained drone operators who are able to stay current in flying, because they go to so many accident sites. Some other AAIBs don’t have that, so it would be down to investigators to get trained, or have nominated investigators, which could be a drain on their resources.
I think people see the outputs and recognize that they’re useful. It’s just added utility: it saves time not having to go around and GPS map each point, take lots of tape based measurements between points, or manually draw a sketch on paper.