Thanks guys for your esfford to protect this incredible animals, I wish new technologies and ardupilot helps to win the battle, good luck
No thanks required, this is a war that all those I know who are personally involved feel is an obligation. We are simply trying to save a number of species from our own kind.
Sorry to keep on about this and excuse my ignorance of the landscape but doesn’t the presence of poachers cause measurable changes in the behaviour of animals? For example, in forest, the presence of humans will trigger warning bird calls (and monkey calls if monkeys are around) but in your case, the bush probably doesn’t offer that (or does it? I really don’t know).
Still, the reason I am asking is because I remember a project where mobile phones were mounted on trees to listen for the sounds of chainsaws in order to detect illegal logging. Listening posts are a lot easier than flying platforms, you mount them, you stick a photovoltaic to them, arrange for connectivity (mobile mast, LoRa, mesh network, whatever) and they keep going with minimal interference. Recently, I had spoken to a biologist who was looking into soundscape ecology (a fairly modern discipline). I am wondering if you could leverage any bush noises (or any other measurable indicator) into an early warning system of human contact.
The issue here is not only the terrain, but also the size of the reserves. Poachers are not stupid when it comes to being bush wise and know ho to live in and move around the bush. Its a nice idea but now sure it will be effective.
Of course they are bush wise but there is nothing to do about it. Birds and monkeys (and I 'm sure others as well) will call out a detected threat and there is nothing anyone can do about it, if there was predators, who are just as bush wise as any human, would have figured it out in the evolutionary arms race. The question is, if there are enough species that will call out a disturbance and how far away you can detect it, if at all. If you can, then you have ground sensors for free and what you need to do is build the aggregator network that will bring you their output (mics and wireless, cheap and low maintenance). Then, your drones (and whatever other resources) go on top in a more targeted manner.
With the number of predators you will only know the animals have been startled, especially the buck and you won’t know why. Also unfortunately many animals are used to seeing humans and they don’t necessarily see us as a threat. There are projects looking at monitoring when Rhino panic but by the time they do, it is often too late.
Hmmm, you have a very good point there. You will be drowning in false positives. Another “cheating” way I can think of to get around the fact that computer vision is almost there but not quite there yet is to use animal vision. Just thinking out loud here but would it be possible to train raptors to spot poachers? Their sensors are unparalleled and their cost and maintenance probably comparable to a high end drone (though I don’t really have figures).
In a hybrid mode, you could do something like: if you spot something do three right hand circles or wiggle or whatever gesture that can be picked up by a tiny autopilot board strapped to the bird’s body. Then the location and the trigger event can be transmitted and raise a flag with the ground crew. The bird can stay high up, out of harm’s way and you can send the drone in for confirmation.
The main difficulty I think might be (though I am not in any way an expert) to somehow train for spotting groups of humans and then performing a gesture, both of which are not something that comes naturally, unlike hunting prey, which is what they typically trained to do. Still, perhaps having a chat with a falconer can’t hurt.
I did discuss the falcon idea with some of the wardens I work with and the answer is that there are wild populations of raptors with well defined territory. A tame bird would be open to attack and could also disrupt the natural patterns. The bush is a very complex ecosystem with problems that make protecting it harder.
As a wildlife biologist and researcher, I am investigating the potential of UAV based technology since several years, and I’ve frequently been surprised by the “stubbornness” of approaches relying on thermal camera imagery within the hot climate of (sub-)tropical Africa.
First of all, I mostly appreciate the dedication of people writing on this blog to contribute to the conservation of nature. We definitely have to improve the methodologies and technologies and make them available (!!!) to the managers and rangers of protected areas.
I am currently working on a recently launched social-ecological research project investigating (among others) the efficiency of protected areas for biodiversity conservation, from the governance (social) as well as from the ecological perspective.
Being responsible for the ecological work package, I’ll finally have some funds and possibilities to take a closer look at how to effectively and sustainably integrate UAV technology in wildlife management in African protected areas.
I thereby focus on VTOL-capable airframes, and I’m happy realize that other projects follow the same approach. I see the advantages of such a system design not only in the potential to take-off and land on small open patches, but also in the high degree of procedural autonomy (relieve requirements of piloting skills, leading to a real TOOL for non-UAV-specialists) and the protection of the sensory equipment.
What we found so far regarding the technology and its application within the African context, is
a) the repeated failure of reliable use of imagery systems (mostly relying on thermal imagery) - I could add here a long list of (more or less failed) projects trying to detect mammals (incl. humans) -,
b) a general lack of proven protocols on operations involving UAV based data collection, and
c) the high systems costs which hinder the accessibility of the technology to a wider audience (conservation institutions, PA managers, etc).
Our idea is to investigate and develop a system which:
i) relies on imagery principles that ensure the object (in our case medium to large-sized mammals, as we focus on automatised wildlife monitoring) detection INDEPENDENT of environmental TEMPERATURE
ii) can operate highly autonomously
iii) is affordable to individual or pools of PA managers
iv) needs to be robust and easily maintained with simple tools
Although, I realize that the detection of poachers is a different story and might require a different way of thinking (as the behavior of poachers is different to the animal behavior), there are multiple similarities. E.g. operational protocols need to be reviewed and potentially revised in accordance to the flight performance of the UAVs, as these systems are somehow (still) limited in endurance and/or speed - let’s forget for the instant about the issue of legal constrains. I further had to realize that funds for projects related to the conservation of highly endangered flag-ship species is much more accessible and extensive, and costs are probably much less of a concern, but I also think that anti-poaching operations need to be available in a long term and a wide audience (i.e. not restricted to PAs with high reputation, tourism income, etc.) to have a sustainable effect.
As such, I have recently applied at the National Geographic Society for further funds to enlarge the scope of imagery sensors working with ALTERNATIVE features of light and vision, and I am impatiently waiting for their feedback.
As such, I am most interested in your proceedings and happy to discuss related issues with all UAV experts here.
I agree wholeheartedly with your sentiments and comments. I’ll reply in more detail later but we took a ground up approach to build a UAV that is built to address the issues, many that you have correctly raised.
Our current air-frame was designed in conjuction with an old school, aerodynamicist. Working with him was an eye opener and an absolute pleasure. From the beginning he made it very clear that you start with the maths and then you use all the fancy simulations to confirm the maths.
The starting point is what do you want to the UAV to be able to do. Must it fly fast or must if be able to fly slow or does it need to be able to do both? If it needs to do both then efficiency is an issue. If you want high efficiency, choose one of the two options. What is the maximum takeoff weight? If it is a slow flyer and it exceeds the designed takeoff weight then you throw efficiency out the window.
Do you need it to have VTOL capabilities? If its VTOL you have to design it around efficiency because efficient lift motors are not necessarily efficient when flying horizontally.
We spent a long time going through all the requirements, making hard decisions and eventually we decided on a slow flyer that is highly efficient, has VTOL capabilities and has a takeoff weight of 15kg.
Cameras alone are not going to work as they are implemented now and we are looking at new ways to use camera system technologies by combining them. We are also looking to combine cameras with other sensory technologies.
Your 4 points are all factors that were taken into consideration right from the start, especially affordability and maintenance.
Our idea is to fund raise the builds of these UAVs so that its not only wealthy private reserves that can afford them.
Are you guys aware of aerotenna? They had made a splash in the news a while back but no idea what happened since. They seem to develop radar hardware for UAVs, as well as a cool looking, Ardupilot capable board. It could be that their radar sensors might be of some help.
If anyone does use their stuff, could you please tell us how they perform? I, for one, am very curious about them.
Thanks, we did come across their units but their radar will not be able to detect a human. What we have found is that the Sony Stravis sensors with 4.6 million ISO (stealth marketing at its best) are potentially able to provide clear colour video (similar to the x27 Osprey from SPI) at night with only starlight illumination. We have a unit on the way to us to to start testing. If this works the combination of this camera with Thermal will mean a huge jump forward for us.
Hi Robert. Would you mind sending me an email ? Would like to chat offline.
I have been a little quite lately because we hit a minor speed bump being that our front motor placement induced significant torsion stress on the inner wing structure. To overcome this we designed a totally new wing structure that pushed Solidworks to its limited and chewed up over 2000 hours of computational time. I am happy to say that the test section of the new core drastically surpassed our expectations not only in lightness but also in strength.
The good news is that we can potentially lighten the core even more than we first thought.
This is the core segment without its carbon skin.
Thanks for posting the pic. Was your section printed with PLA? What are the dimensions and weight of that section (hard to tell from the pic)? Why are you covering w carbon and not just 3mm packing tape? I assume you’ve seen this fabrication technique here https://www.3dprintedrcplanes.com/kodo/? He uses a cf spar.
The wing section was printed SLS using nylon 6 but the final wing will be printed SLS using Carbonmide. The full wing is 3.3m and the section above says we are on track for a weight of 2.9kg. The wing has to lift a 15kg UAV with a significant torsion force from the front rotors so the carbon skin was designed to add additional stiffness and structural integrity. What you can’t see in the picture is the capped sine-wave spar.
It’s been a while since the last updated but we are still moving forward. All of the CAD work and simulations are complete and we are a week away from the first flight trials of the halfscale airframe.
Hi, how much does the scale prototype weigh?
1.5kg with servos, receiver and batteries. It is pretty much a solid foam constriction