Editor’s Note: This is the second part of a two-part post about using drone technology to search for orangutans around the Sikundur research station in North Sumatra. To read the first part, click here. For all posts in the series, "Call of the Orangutan," click here.

After our drone, which was designed to help our team scan the forest for orangutan nests, went missing on its way back from a scouting mission, a week went by during which the field staff went out every day searching the grid, but there was no sign of the missing unmanned aerial vehicle (UAV). So, Graham Usher, the landscape specialist at the Sumatran Orangutan Conservation Programme (SOCP), came back with a new plan: to search for the missing drone with another one (again the Skywalker type). This time the weather was perfect, and the small plane was soon prepped and ready for flight.

Despite the presence of a number of Swiss interns working for SOCP’s founding organization Pan-Eco, I was sent out to launch. Again, I was nervous as hell, but having done it successfully once, at least I knew what to expect. After the previous week’s near miss with the trees, the launch point was shifted slightly so I was throwing down a dirt track with some nice ferns on either side that would provide a soft landing if anything went wrong. This time, the weight and strength of the drone wasn’t as much of a surprise, and I managed to get it out of my hands and airborne straightaway. Quickly, the drone set off on a 20-minute flight that would take it over the area in which the previous mission was lost.

The time passed quickly as we crowded around our flight control monitor watching carefully to make sure a similar fate didn’t befall the second drone. Before long, it was back, automatically circling the location from which we launched with more than 50 percent of the battery left to play with for the landing. My second task as Graham’s “assistant” was to gradually bring the UAV low enough for him to land it manually, which I did by reducing the autopilot’s altitude 20 meters at a time. Once the drone was around 50 meters above the ground, he switched to manual control and brought it in to land with a gentle thump (and general celebration from all watching).

Minutes later, Graham had the still photos from the downward facing GoPro camera up on the monitor, and we eagerly scanned them searching for the lost drone among the trees. Amazingly, it was easy to spot. It sat neatly on top of the canopy in a position we probably never would have seen during searches from the ground below. Graham used the camera’s clock and distance to work out accurate GPS coordinates and found a large dead tree to use as a marker. He then sent the field staff out into the jungle to recover the ill-fated drone.

While the field staff went off on their mission Graham set up a second UAV, which would carry a larger Canon Cybershot camera, set to take a photo every five seconds in order to get the higher resolution required to conduct an orangutan nest survey. The drone was a MAJA, a bigger airframe, which looked very boxy, and not the most aerodynamic piece of kit I’d ever seen. However, it apparently flies decently and has a huge storage compartment allowing for up to two kilograms of weight to be carried as a payload.

Weight is basically the limiting factor in what applications the different UAV frames can be used for. Once you’ve factored in the autopilot, battery and receiver for remotely controlling the unit, with a smaller frame there isn’t much capacity left, so the bigger MAJA has an advantage in that respect. However, this extra weight reduces its flying time and also made it much more difficult for me to lift, so coming up to launch, I was a bit worried about whether I’d be able to keep it horizontal. However, we were running out of time to get the three runs required to do our 10 transects, so I didn’t really have much chance to be nervous before the preflight checks were complete, and I stood on the dirt track ready to go.

Graham warned me the MAJA would “pull harder because of its bigger engine,” which was definitely true. I almost lost my grip and struggled with the big frame. Once again, I took five fast steps and threw. It felt like a good launch, with the drone only wobbling slightly to level. But then the model plane, seemingly fine, just lazily barrel-rolled left and plowed straight into the ferns lining the track!

It was disappointing. Actually, I felt terrible about it even though everyone assured me both the launch and frame were fine (apparently they can take a much harder battering than a crash from 3 meters up). So after some minor repairs, I ran, threw and once again the plane rolled and stuck itself into exactly the same patch of ferns. In all, I tried 4 launches, my arm gradually becoming more and more tired and my throws more and more ragged, until I was unceremoniously hooked from the mound.

“Maybe someone taller should have a go,” someone said, so SOCP’s director of biodiversity monitoring, Matthew Nowak, stepped in. As a veteran of six launches (two successful), I feel able to offer insightful comment and analysis on Matt’s efforts: He is taller than me, which definitely helps, is much faster than I have ever been, and has a bigger frame allowing him to hold the UAV up with ease. All of these factors contributed to what I would describe as a textbook launch.

Yet MAJA once again plowed into the ground in exactly the same place, indicating some kind of mechanical problem, and we decided to abort for the day.

Spirits were low, due to our continued failure, but we were all cheered by the triumphant return of the field staff carrying the original UAV from a week earlier. Using the coordinates and markers, they’d quickly located the downed drone 25 meters up in the canopy. Fortunately it was also next to a steep hill, which they climbed, and using a long stick they poked the drone free.

So why had it crashed? It turns out the final waypoint above our launch location was set to 90 meters (this should have been high enough, but there is a lot of variation in the sensors and topography), so the drone tried to drop down to this height immediately after completing its final turn, cruising through the canopy briefly before nestling (fairly) gently in the trees. The frame is still in great shape, and although it sat in wet conditions for over a week, one of the two GoPro cameras is still working, and Graham managed to salvage the data off the other.

We’ll retrofit the drone to carry the Canon camera, and finally complete the orangutan nest surveys next week!

All in all, for me, it was a challenging, but rewarding experience helping out with the survey effort. Although the drones look like toys, they actually require a huge amount of equipment and technical knowledge just to be able to fly them, let alone deal with the complex survey methods required for doing science. This is one of the most exciting projects I’ve been lucky enough to work with, and using drones is only going to improve our ability to collect vital, high-resolution survey and monitoring data at a vastly reduced cost—not only for orangutan research in Indonesia, but across a range of taxa and habitats.