To borrow a title from a former chancellor of our alma mater; in this post we'll provide a paraphrased history of the project from inception to the present day.
Intro and early days
Getting straight into it - the idea for the project came from a conversation between Tom M & Tom A at a friends' wedding in 2014 - thanks Andy and Michelle!
During the project we've been very fortunate to be invited to the Natural History Museum to give a talk and have a stand at a couple of events. One of the questions that comes up most often is "what company or university are you at?" We're not. The project is simply a very enjoyable hobby. We want to help further bat conservation it's a great way that we can assist with this using our respective skills. Tom A is our bat expert (and is now a qualified and competent drone pilot) and Tom M is our drone designer, builder and operator.
We think this goes to show that if people are willing, you don't need the support of a big research grant or institution to push boundaries. Our advice is to "give it a go and see where it takes you".
We knew that quadcopters produced a lot of sound but didn't know how much of this was ultrasound. Our first tests showed us that it was a considerable amount. This ultrasonic interference has been a running theme through the project. The problem lies in the signal to noise ratio of the bat call. If there is too much noise from the drone we won't be able to 'hear' the bat call.
This meant that we had to separate the drone and the detector recording the bat calls by about five meters. The highly technical application of a piece of string was employed though this wasn't without it's problems as you can see in the video below.
The problem of the oscillation arose due to the high weight of the detector. We were glad we used a water bottle of the same weight as the detector in testing! The detector and recorder weighed about a third of the quadcopter mass. This led to loss of control and meant and it wasn't an option in this configuration.
In going back to the drawing board we wanted something that produced less noise and could comfortably fly with the payload of the detector. The obvious choice for us was to move to a plane; it had the added bonus of being able to fly for longer too (about 8 mins for the quad and 25 for the plane).
The Talon is one of our two current options for flying bat drones (the other being the quad, see here). It's a small plane which, like the quad, can carry a small bat detector and acoustically record bat calls. Have a look at the first post on it here for more info.
One of the major issues we've faced with acoustic monitoring is noise from the propellers. We found that if the detector is not carefully positioned a certain distance from the propeller/s then the noise may drown out any bat calls. This is the latest post in a long line about noise, see the rest of them here.
Last blog post about the Talon we looked at a number of different positions of the detector and discovered that about 15cm from the nose is the quietest position we tested. This led to us considering an extended option for the Talon - which is to say, having a stick pointing forwards off the front of the aircraft that the detector is attached to.
And it looks like this:
We've secured a glass fibre tube from a fishing rod onto the front of the plane. This increases the distance between the detector and the propeller (the source of the noise). As it's ultrasound, it attenuates quickly so a small increase in distance can lead to a large drop in noise.
Oddly, we found during the initial noise testing that the closer position (5cm from the nose) was actually quieter than the further position (15cm from the nose). We're not sure why and would hypothesise it's something to do with the airframe blocking the noise at 5cm.
The next stage of testing performed was actually flying with this setup and seeing how our two detectors perform (the Peersonic and the AudioMoth). For interest, we tested the noise levels of the two detectors at two distances from the nose of the plane, 15cm and 50cm (the furthest possible distance).
Here's how the noise testing turned out:
There is little difference between the two distances (although the 15cm does seem to be a bit quieter) but there is a distinct difference between the two detectors. The AudioMoth is particularly sensitive to moving air across the microphone resulting in extreme wind noise.
The noise levels on the Peersonic are significant but limited to a fairly narrow frequency which will still allow us to detect a number of bat species easily. It is also worth noting that the amount of noise varied through the flight, some louder and some quieter than the examples above.
We're looking forward to getting into the field for some real world testing with this setup which we'll be doing next!
After our initial field testing with the first bat plane (have a look at the blog post here) we knew that we needed to reduce the noise seen in the sonograms. This has been the driving force behind making the Mk2, Mk3 and purchasing the Talon.
The noise is a product of the propeller spinning and as such we've tried to reduce the noise heard by the ultrasound detector by separating the propeller and detector by as much distance as possible.
For the Talon we tested both the Peersonic and Soundtrap (which has changed name to Open Acoustic Devices due to a conflict of names - the actual detector is called the AudioMoth) detectors in a variety of positions to compare noise levels. The positions were:
* This would be achieved by attaching an extension to the front of the plane. For fun we've called this the Nimrod option. It'll look something like the re-fueling pipe extending from the front of the plane:
Let's take a look at the data we produced from the two detectors in the various positions. For reference, the throttle was set at the minimum value required for level flight (circa 35%) in order to minimise noise which mimics what we would do in field testing.
Here are the sonograms:
The first noticable difference between the two detectors is that the Soundtrap picks up a lot more noise. There are a number of possible reasons for this. One we know will be a factor is that the Peersonic has a microphone which actively suppresses noise below 20kHz.
Looking at the difference between the various positions, it's apparent that the noisiest position is under the belly (also closest to the propeller) and the quietest is 5cm from the nose. The quietest is somewhat of a surprise, we expected the 15cm from the nose position to be quieter as it is further from the propeller. Perhaps this is due to the airframe blocking the sound at the 5cm position.
In conclusion, the current thinking is to use the 5cm position for field testing. As there is still some time before the bats come out of hibernation we think it might be worth testing a further extended-Nimrod option, with the microphone circa 60cm from the nose and we'd also like to compare the signal to noise ratio for the Peersonic and Soundtrap (AudioMoth) in the preferred position using the ultrasound source (see the blog post here about using this previously).
Keep checking out the blog for more!
The last time we flew the plane we pretty much smashed up everything but there was just enough left to test a new set of propellers that we got.
There's no photo of the DJI q-tip prop unfortunately but if you imagine the DJI prop with about 5mm of the tips turned 90 degrees into the flow of the air then this is what it looks like!
We tested each propeller for the amount of ultrasound that it gave off and here are the results. The axis along the bottom is the frequency and up the side is the volume, so we want the blade with the lowest peak.
Top left - clockwise: 3 blade, DJI normal, DJI with q-tip, Graupner eprop, below: Graupner slow fly
The Graupner slow fly propeller, above, (which is wider than the other propellers) was much quieter than the other props. In fact, since the volume is measured on a log scale (decibels) this propeller was 10 times quieter than the loudest prop - the three-bladed propeller.
This test demonstrates that the propeller in use does have a significant affect on the amount of ultrasound produced and that is is possible to reduce the ultrasound interference from this source.
In our efforts to replicate all of Batman's modes of transport in miniature we thought the Batboat would be a fairly easy hit...
This was an idea which originally came about from discussing the possibility of monitoring bats over water using the BatUAV. We then thought, why not just use a boat? This idea was put aside as we were concentrating on the planes but later on Peter at Peersonic also had the idea independently and put it to us.
The result is this fine vessel:
Oh wait, no, that's not the one...
This looks more like it, the little one called Princess:
So we've ended up with a boat from hobbyking called Princess.
She's about 1m long and 30cm wide and can travel incredibly quickly (at about 35mph, which we don't need but it's always fun).
There's quite a few challenges to overcome in using a boat autonomously:
We've plugged in a Pixhawk flight controller which can run Ardurover which controls ground vehicles such as a boat. The Batboat works just fine using this. Have a look at our video below for the initial runs.
As for the ultrasonic interference we've tried out a new microphone from Peter which discriminates against 8kHz and below (to try and cut out some interference). Below are some sonograms of the results at the bow and the stern.
Obviously, the interference at the stern makes the microphone here essentially unusable. The bow is much better but we'll need to protect the microphone from any splashes. Hopefully we'll be able to turn off the motor and ensure there's practically zero interference whilst collecting data. Exactly how this is going to work in moving water is uncertain at the moment as we'll need to test position hold.
We didn't manage to fit in any GPS testing (ie autonomous piloting) but watch this space for more testing soon!
Following on from our post here about our new detector from PeerSonic we went ahead and did some testing with this on the plane.
We wanted to ascertain how the plane/detector performed for two criteria:
For a quick reminder, here's the set up we were using as modelled by Tom A:
The position of the detector on the wing tip is not ideal. It leaves the detector open to damage upon landings and adds additional weight at an extreme position relative to the centre of gravity which must be counterbalanced and gives rise to instability. For flight testing, we balanced the weight of the detector with a bag of sugar on the opposite wing.
We're currently working with Peter at PeerSonic for a solution to this where we'll hopefully have the microphone separate from the detector and be able to mount the detector more centrally (relative to the centre of gravity for the plane).
So, how did it fly? There was a noticeable instability in the roll element of the flight but nothing that the flight controller stabilisation and pilot input couldn't correct. This was more apparent when windy but the plane was definitely flyable in this configuration.
For the ultrasound detection testing we flew two passes at roughly 10 m from the detector horizontally and at altitudes of approx 10 m and 20 m. Here are the sonograms from these passes:
The ultrasound source is easily visible at 10 m, especially as it's a different frequency to the propeller interference. At 20 m we can see that the ultrasound source is hardly detectable.
Fortunately, we expect bats to be somewhat louder than our ultrasound source and hope to be able to detect at 20 m and perhaps even further.
So, great news! We now have a working prototype for the Bat UAV! In the next post, we'll be doing some actual field testing and see if we can record any bats!
We've had some success recording an ultrasound source whilst in flight with our current detector set up that can be seen here.
Unfortunately, this set up does not enable recording whilst the motor is in use and required the gliding flight that we tested this with (see here). Our ideal solution is a detector that can record whilst in normal level flight and we might just have been sent one such detector!
Having posted our blog updates in the UK Bat Workers group on Facebook we had a reply to one of the posts from Peter at peersonic.co.uk. Along with a colleague, Peter has developed a very light and compact detector:
Peter very kindly offered to send us one of his prototypes and we were very happy to accept and give it a go.
Having thought about the placement of the detector we decided that if we placed it at the end of the wing we might just be able to get away with detecting the bats over the noise of the motor.
We came up with a way of attaching the detector to the plane (though at this point this hasn't been flight tested). The photos below show how we've attached the detector and put a polystyrene protector over it.
With our new set up we wanted to test whether an ultrasound source can be detected over the noise from the motor 'in-flight' . To do this we held the plane steady approx 2 m from the ground, set the throttle to circa 40% (which should be roughly level flight) and turned on our ultrasound source at multiples of 2 m from the detector.
The analysis of the data was then done in two parts. Firstly, Tom put together our usual graphs using R which produced the graphs below. It's worth noting that all the graphs use a negative decibel scale. 0 dB is the loudest sound detected and everything else is measured relative to this sound.
The blips that show up regularly at about 30-45 kHz are from the ultrasound source. The lower frequency noise between about 20-30 kHz is from the propeller and motor.
Using this software it looks like we can detect the ultrasound source up to about 16m.
However, when the data was run through BatSound we were able to extract a lot more useful information. Have a look at the graph below which shows the distance at 20m and you can clearly see the blips from the ultrasound source:
So, we've got some good news: using the BatSound software we're able to detect up at 20 m and potentially a little further. We suspect that bats are going to be somewhat louder than our ultrasound source but would caution that ultrasound attenuates quickly. We'd be very happy if we could still detect at 30 m but we'll have to do some more testing to see whether this is the case.
As an interesting aside the graph above covers about 200 ms, the motor/propeller 'noise' (~20-30 kHz) is now split up into the actual rotations of the propeller. Working through this, we can tell the propeller is rotating at around 3000 rpm.
The next test we'd like to perform is a level flight over the ultrasound source at varying heights so watch this space. If we can achieve a recording from this then we'll progress to attempting to record some real-life bats!
We'd like to take this opportunity to give a big thanks to Peter at PeerSonic who kindly provided the prototype detector that made this all possible. Thank you!
It's all coming together! We have a rig that allows us to attach an ultrasound detector and recorder to a UAV (have a look here on how we did this), and we can use gliding flight to record without interference from the engine. So we gave it a go:
Flying the drone around on a loop we used a ultrasound source to produce noises that the on board recorders would pick up. The results are promising. When flying overhead at around 20 meters the detector was able to pick up the ultrasound signals emitted from ground-level. Since bats are often flying at tree height this gives us a good chance of picking them up in flight.
The glide slope we experienced was approximately 3.5:1, meaning that for every 1 meter altitude lost we flew 3.5 meters laterally. This isn't particularly good and the glide slope was reduced by the fact we were flying into the wind whilst gliding. Although this provides additional lift, the reduced ground speed has a large impact on the slope. In future flights we'll fly the glides downwind as much as possible.
The next step will be to plan and fly an autopilot mission using the gliding method to try and record an ultrasound source.
It's worth noting that we don't think this is the ideal solution and would much prefer to be able to record all the time so we're looking at other options for this.
When we tested the quadcopter we found it produced a lot of ultrasound interference. This time we tested our new plane (Bix-3, Hobby King) to see how it compared. We had three different propellers to test, and we also tried out some acoustic foam to try and dampened the noise from the engine.
In the graph you can see each of the propellors types across the top and the results with and without foam along the side. The three lines in each panel show three distances from the nose of the plane, 0cm, 50cm, and 100cm, the top line being the closest (0cm).
You can clearly see that across the three propeller types the foam makes a big difference to the volume of interference (note that y-axis has a log scale). All the top panels (with foam) have a lower amount of interference than the paired bottom panels (without foam). When we compare across the propeller types the Bix3 clearly outperforms the others, and the volume levels at 50cm with foam is similar to that at 300cm with the quadcopter, a great improvement.
So, as we know from the experiment with measuring the ultrasonic noise output of the quadcopter at different distances that the microphone needs to be at least 3m away from the quadcopter.
As shown from the string suspended weight this produces stability problems when the 'detector' swings about.
There seem to be a number of solutions to these problems:
What we're actually going to try:
All of them! Might as well try everything as only by experimentation are we likely to find the best option.
Firstly, we've bought a plane and will be doing the same noise test with the propeller interference. We'll also be trying wooden propellers to reduce this. The plane is a Bix3 from Hobbyking:
Secondly, we're trying to reduce the weight of the recording equipment. Instead of using a bat detector and digital recorder we'll try using this configuration:
Happily, we can use a mobile phone instead of a tablet to help keep the weight down. The microphone is an ultramic from Dodotronics.
Thirdly, we'll attempt to mount a lightweight (~100g) carbon fibre rod below the quad which will pivot with changes in pitch. This may provide problems with the roll of the quad but we'll have to see.
Finally, we will be shielding any microphone with noise insulation foam.
Hopefully through the combination of these efforts we'll have a viable vehicle to do some real world testing!