Drones: Tag Tracking Project | Duke Conservation Tech



hey my name is Peter Moran and I'm working on the drone team at Duke conservation technology so the main thing with our project is we're trying to track animals using a drone giving researchers the location of animals they can help understand them better and understand their behaviors better now for us we're looking at turtles so you could put a GPS tracker on them but if you wanted to say track a hundred of them that would require a hundred these trackers and that can become costly and eventually they fall off and need to be replaced so instead by using these tags which are really just a convenient identifier for this camera to see that the computers are really good at finding there's only a few cents for your tag when they fall off you can just replace them we actually have a few different kind of components going on first of all we got this kind of this whole drone here one of the cool things about it is this here is actually a computer and this is also what controls the flight and that makes it really easy to work with and easy for everyone to develop on but we also have this this little cube here and this is essentially a supercomputer so here is specifically this is called the Nvidia TK one and here is the main thing under this fan is the processor and it's like a specialized processor that's really good at doing stuff like computer vision and because this is so small because there's so few parts on it it uses low power and we can power it off a battery this is our camera that we bought specialized for computer vision it is very tiny but it actually gives us 4k video and it's nice and easy to use with the USB so we got this computer and so we can just plug these cameras into and then you take this ethernet cord and you can plug that in here and and now we're connected to the drone but for the computer vision really what's important is these tags so the computer actually knows how to look for these in every frame it's looking for squares and then it checks like is it they're kind of encoded square the way up here and since it knows the size of this square it can figure out the distance away from the camera can figure out the sizes of things around it and then all this in here this is essentially the part that's like a QR code so all look different and it represents a number and so we can come track each animal as an individual and understand how are they interacting we thought each other and it's actually doing all this tracking live fuel and it's very when it's flying you can see it that also lets us do things like track animals and you can do that live so right now we've essentially gotten the basics of flying around seeing these tags and logging the GPS location of our drone now the one thing that we're kind of looking to add on top of that is we can identify these tags but also the way they are because we know the size of them and because we know they're kind of original shape you can actually get the position of these tags relative to the camera and produce anything more accurate estimate of the turtles location itself beyond just kind of like using this drone for other applications we'd like to make it work more easily for the researchers themselves so most likely would make an app that could connect to the drone and the drone can just pay attention and look for commands so there's a lot of things like that well we want to add there's really a lot of different things you can do by tracking these animals and we're just trying to do this in a cheaper way and I think really it's just interesting overall because this not only can it be applied to the Turtles but we think this is kind of like a platform for DCT to use for a lot of different remote sensing applications there's also been a lot of interest from some of our contacts about sea lions that's one of the cases where they're interested in tracking the spread of disease sea lions are not gonna be as easy to track as Turtles because you can't just have one side that's guaranteed to be on all the time so that is something that we saw the think about and how he wants it to our future plans like I said is especially get this to work more reliably we're actually upgrading on computer because that will make it easier to work with and it's like four times as powerful which means we can get better data well really hope is to make something that actually gets used I think this is kind of common across all of DCT and maybe doesn't always get mentioned but we're looking at projects that we think are of actual value the real impact that we're looking for is to get this in the hands of researchers and for them to use it so kind of looking forward maybe we'll be working with the Marine Lab to integrate into their systems and then they can use it for their own research and that would be really amazing again is to get these people to actually use it and have this real world impact you

Leave a Reply

Your email address will not be published. Required fields are marked *