The Devils in the Data, Wearables for Wildlife

Most of the RPAS flying I do is in support of conservation projects in Africa, its in fact where my unmanned aviation interest came from. The longest trial to date a radio collar tracking project. The subject animals are mainly cats serval and caracal with a smattering of jackal. We are also involved with the tracking of Rhinos. I can mention them in this as the five we recently collared were all over the press here in South Africa.

What I want to try and convey in this missive is that its not as simple as folks think.

Duel frequency UHF/VHF rhino collar, it is attached to their legs.
Duel frequency UHF/VHF rhino collar, it is attached to their legs.

The hype that I see generated by vendors parachuting into Africa, selling what they claim to be able to do, and then running away fast is just amazing.

Several demonstration videos of wildlife being spotted from above are of animals stood out in the open.

The world is not flat and entirely made of grass.

The reality is that Africa is a rugged place, mostly high altitude and hot. Platforms that are a little overweight in Europe just don’t fly on a summers day here. In the wildlife community RPAS are beginning to get a bad name.

Money that could have been spent on real research or other countermeasures is being wasted on systems that don’t deliver.

Of course I need to be part of the mission to change that view.

twoserval
These were two rescued serval that we released to the wild.

Two years ago I was very lucky to be contacted by Victor Hugo. He is a fixed wing pilot of some experience and also into model aircraft. His day job is to create radio collars for wildlife research projects and asset management. He could see the potential to create virtual towers to extend the received range of his repeater units using model aircraft to lift them. A more passionate wildlife advocate you will be hard pressed to meet.

He asked me to help. We are on our way to 300 flights in support of the task the repeater pulling signals in from up to 20km when conditions are right.

There are many types of collars from many manufacturers around the world already placed on animals in Africa.

Victor fits two types, an older VHF pinger that does little more than emit a signal every few milliseconds that a researcher or ranger swings a beam towards to determine a rough direction and then change their position and try and triangulate where the animal is. They are cheap.

The second more modern type operate on UHF, have a GPS, temperature and movement sensors onboard. When in range of a base station this type not only sends its current position but all the data recorded on the collar since it was last seen by a base station.

Having base stations on hilltops listening out works very well indeed. They do from time to time go down for tech reasons or the study animals start living over the next hill. This is where the RPAS comes in to close the gaps.

Just how well the system can work unfolded recently.

The very first serval that we collared early in 2013 disappeared from our system after five months of tracking. The last GPS position was close to a road and I flew and walked and drove all around that point hoping to hear the collar. We could see there was quite a bit of data stored on the collar but needed to get closer with a base station that would pull the stored information.

No new GPS fixes were coming, if the collar is lying upside down you would expect that. If you have a collar trying to send all its data and dropping packets it starts again and will quickly flatten its battery so they don’t only being will to talk when they know the entire message will get through.

Fields either side of the road next to the last position had been freshly ploughed and in the end what I believed to have happened was that the cat was hit by a car and made it into one of the fields and was perhaps ploughed in. After harvest I would come back and check again. Harvest came and went and nothing nada.

Imagine our surprise when at the beginning of October this year one year after it went quiet and getting on for 18 months since fitting the collar it popped up on the system again. This time with a fairly strong signal.

Victor was straight on the phone when he saw it pop up and I gathered handheld radios my son, his friend and scrambled to go out and look. We had no idea how long the collar would stay trying to report. Straight to the last known position we rushed and yes the fields had been worked, maybe they had dragged it to the surface. No GPS position but a signal.

Collars normally transmit every 10 minutes looking to hand shake with a base station. Hearing that signal is hit and miss as its weak. But eventually with two small boys spaced out along the road each of us could hear the burst of data on our own handheld. This was enough evidence to command the base station to change the mode of the collar. We had to be close This makes it transmit a couple of times a second, just a blip. Allows us to go old school and swing a beam to look. This action flattens the battery fast.

Victor made the command happen and watched what the collar did on his computer at home. We waited for the collar to change mode. It didn’t.

Daylight was going and there was not enough time to fetch the base station down into the valley to try and get it closer to trigger the command with a stronger signal.

At least by not changing mode we were not flattening the collar battery at a faster rate and we knew we could hear something every 10 minutes at least.

Time to go home, charge the equipment and ready some new solutions that we have been playing with.

We can now map received signal strength and that really helps with legacy collars with no GPS. I am in still in the process of making the equipment smaller in order to fly it and am currently delayed by the post office strike that is happening in South Africa. Its nearly four months old and that’s where a few key electronic components are sat. In the meantime I have the software running on a laptop tethering my cell phone for a GPS.

At 02:00 in the morning for whatever reason the signal between the base station and collar strengthened for long enough for the collar to receive the command to talk faster.

Now it was a race against time.

My son and I were on scene at 0800 and recording signal strengths on the laptop. Within an hour we had narrowed the search down, close to a very small dam. We then walked many times around the dam and looked. Noting where the strongest signal was. Victor was in his car and on the way with his beam. My son and I were just not finding the collar.

When Victor arrived I tried not to prejudice his search with our opinion, we had got him somewhere within 50 m of the collar.

We moved the base station down from the top of the hill and were able to pull all the data the collar held.

Eventually after much head scratching Victor realized the collar must be in the dam.

To confirm this we sent a willing volunteer (ok told my 10 year old son it was his job) to wade into the water, reducing the sensitivity of the receiver as he went to narrow down even further the position.

IMG_20141024_115619

Within the fence line running along the dam wall we had found a couple of snares and that triggered worry in ourselves.

On analyzing the data at home the accelerometers filled in more detail.

The activity level of the cat goes through the roof, above a point that would trigger an alarm if it was in full sight of the base station. For four hours the serval struggles until it eventually dies in the snare. Activity drops to zero. Several days later the collar accelerates again briefly when the snare is checked by whoever set it, the cat removed and the collar is thrown into the dam.

What we think happened now is that the collar reported in as normal and that was our last position report the cat then moved in the shadow of a spur so the base station could not hear it. Perhaps it was going for drink or to hunt for rodents at the edge of the dam but it was there it was snared.

The reason it just popped up again? We have been experiencing drought and the level of the dam had dropped enough for the RF to get through.

deathofaserval

A serval skin would fetch a tidy sum of money and might be used in traditional medicine or a part of a costume if intact.

The sum of parts and people required to conduct real wildlife research with unmanned aircraft is much greater than just the purchase of a system from the most convincing salesman.

budgies
Scene of the crime from Budgies DJI Phantom

Sadly the RPAS has proved most useful in finding snared animals, twice now I have been able to hear collars lost by the base stations and luckily the GPS has still been working on the collar.

Its never a pleasant thing going to find them and the story of this one is for the pub really. There were serval4many snares in the fence line and quite a story about the game guard that patrolled the fence line. He was not involved but threatened with not only his life but his families as well if he said anything.

The carcass of this serval had been dragged by other predators to be eaten. The collar was with the rest of the remains. Less than two metres from this cat was a jackal also snared.

A very distressing spot and one we don’t really wish to visit again.

Not only are we and the researchers upset but the landowners in which the study animals were originally caught become heavily invested in the travels of these animals. One positive is that they have doubled their efforts to patrol fence lines removing snares.