WWF's Conservation Innovation Awards

Gathering votes
Gathering votes

Collaborative high tech solutions to make predator elimination thousands of times more efficient

Modern information technologies are a source of lots of ways to improve predator elimination as they tend to get either twice as good or half the price every 18 months.

Open source is the most effective way to get hundreds of technologists using their skills to contribute to a solution that is often more powerful and faster than any one company doing it (eg Wikipedia vrs Encyclopedia Britannica). To date the project has

  • Created a very inexpensive device to monitor the number of birds and to know objectively if it is getting better or worse.
  • Created a heat camera that is more sensitive than any other for NZ predators (most cameras are designed for pigs and deer etc)
  • Shown that artificial intelligence can be used to automatically identify predators
  • The camera can move to follow predators with the end aim or eliminating them (most likely with poison paintball)
  • Use sound lures that act over long distance and are adaptable 

See a short summary of our project here https://vimeo.com/224005468

What conservation problem are you trying to solve?

Removal of all predators from New Zealand. Our solutions will be particularly effective for the hardest part of predator eradication which are the very last ones or for a re-invasion. In those two scenarios food based lures don’t work very well as there are lots of other food sources around but social lures become more effective as they need to have social interactions to breed. We envision lines of these devices moving over tracks of New Zealand eliminating ALL predators as they go.

How are you going to solve this conservation problem?

We will make trapping thousands of times more efficient by

  • Using sound lures that last 20 times longer than food lures, being electrically powered
  • Only require 1/100th the number of traditional traps, as they can operate over ten times the distance (100 times the area).
  • One trap could target any pest: possum, rat, stoat, and feral cats. Hence ¼ the number of traps required.
  • Kill percentage could be closer to 100% rather than less than 10% for many current traps.

These collective improvements mean theoretically the traps could be 80,000 times more efficient: 20 (lure life) x 100 (trap intensity) x 4 (one trap, four pests) x 10 (kill ratio).

See the TEDx talk on the Cacophony project here https://www.youtube.com/watch?v=n9PKUMHDbpw

What makes your idea new and unique?

The two key things that make our project novel are:

-the open source nature of the project that allows anyone to use or contribute to the project

-the focus on digital technologies that get twice as good or half the price every 18 months which means we thing we can make NZ predator free by 2040 https://cacophony.org.nz/2050-target-predator...ems-more-likely

Our novel application of open source tools has been recognised as a category winner by New Zealand Open source.

Our novel approach has also been recognised by financial support from NEXT Foundation, Zero Invasive Predators and Jasmine Social Ventures (Sam Morgan). Spark are providing the project free data upload and data hosting along with expertise and other hardware.

The project has already had significant contributions from a large number of experts and companies that are listed in detail on our website.


Who will use your idea, and how will they benefit?

The project is open source so anyone can use any parts of the project to create products and services for predator monitoring and removal. We have already been in discussion with companies looking to make and sell products and services based on the technology developed. This is a common business model and the example most people may be familiar with is Android operating system that is Open and free and most of the world's smart phone manufactures use it to make and sell a whole range of products and services.

What tasks or activities do you need investment for? How would you spend a $25,000 grant?

The grant will help accelerate the development of the predator elimination part of the project which involves firing poison paintballs at predators. This will build on the toxin work that has already been done by for the Spitfire at Lincoln University. Our goal will be to make the robust tracking and firing device to be triggered when the Artificial Intelligence confirms 100% identification of predators.

See initial prototype of camera tracking here https://cacophony.org.nz/rotating-camera-coul...-more-predators

Are you a New Zealand citizen or resident?


I have read and agree to the Crowdicity Terms of Use, the Conservation Innovation Awards 2017 Supplemental Terms and Conditions, and the Crowdicity Privacy Policy


List five other ideas posted in the challenge that excite you. Why?

Thermal Animial Detection system


Great use of similar modern technologies we are using but instead of targeting rats and stoats they are targeting ungulates. We have been chatting with the folks in Taranaki about using sound lures for goats as well. All our technology is open source so could be used to help this project too


Better way to support permanent forrest in NZ

Love ideas that tackle social structures and these can often have as much or more impact than devices. 


Restoration drone for poluted waterways

Great use of new technology to dramatically scale up how waterways can be restored and monitored


The Drone ranger - areal wildlife tracking

Another example of use of modern new technology to dramatically improve monitoring. Looks like its well advanced with a great team. Love the name too!


Guardians of Zealandia

Good use of intereactive media to highlight what a problem we have and why its so important. When you know this it is pretty hard to ignore so the more people up to speed with this the better. 


Squark Squad

Some great use of technology from a team that has lots of great ideas in this space

How could you improve your idea?

Open source croud platforms are a great way to get improvements in ideas and this is how The Cacophony Project has been running for over 2 years. You can see the details way the project has evolved based on experiments and feedback from over 40 blog posts https://cacophony.org.nz/blog. The input has been from a wide variety of experts https://cacophony.org.nz/cacophony-project-community

edited on Oct 13, 2017 by Grantryan

Brent Beaven Sep 30, 2017

Overall, some great new thinking in this space. Curious as to how you maintain the focus of the open source crowd...will they drift to the next best idea? Some risk elements with firing toxins at predators.....especially in regard to potential risks to other species and people. Why would you fire paintballs, if your proposed sound lures are so effective at getting animals to come in? That said, You have sparked my curiosity in regard to the potential of A.I. in this space. Certainly have been following the bird monitoring work closely...huge potential there for measuring outcomes of PF2050.


Menno Smits Oct 3, 2017

Thanks for your interest and questions.

Regarding your comment about open source, we completely expect that some people will be interested in some small aspect of the project for a short period of time, while others will make more major contributions. That's just the nature of the open source ecosystem and that's fine.

We already have a few core team members working on the project full time and who will keep it going. We also have a number of key contributors and advisers who work on the project regularly. Beyond that, we encourage and coordinate external offers for help whenever they arise.

We completely acknowledge there are risks to firing toxins at predators. These factors will be mitigated by not attempting to shoot at anything until our machine learning model and targeting system are rock solid, and by limiting the effective range (likely 3-5m radius) and angle (below horizontal only). We would also argue that firing a small amount of toxin at specific targets is likely to be safer than some toxin deployment practices currently in use.

Regarding sound lures, these will be used to get predators closer to our devices, extending their effective range. The idea is to get predators that are outside the range of the camera interested so that they come closer. This should allows us to spread our devices out further so that fewer are required.


Gerald Dickinson Oct 3, 2017

Some great ideas here. How are you progressing with the Thermal imaging technology. Looking at your website it sounds like you are progressing well with the Rasberry Pi and FLIR lepton 3 thermal imaging camera. How are you finding the power consumption of these devices? Looking at the data sheet specifications, it does appear to be power hungry and impacting on shorter battery life.
What measures are you investigating to improve the battery life? This is always an issue for in-field monitoring equipment. I have a project in this space and taking a different approach. Let me know how we can help out.



Menno Smits Oct 4, 2017

Thanks for your feedback. In the short term we are not worried about
power consumption but are focused on proving the concept of a device
that can identify all predators over a large area. Until recently the
thermal camera devices we had created covered an area of about 5 by 5
meters (see

We've recently opted for the higher resolution FLIR Lepton 3 camera.
The new camera set up will be on a tracking arm and should be able to
cover over 100 square meters (possibly more). This is obviously more
expensive from a power consumption point of view than our previous
configuration which can only watch a small area but gives a much
better idea of what's going on.

The device will also be able to lure and eventually kill predators
too. Only once we hit this milestone will we focus on optimising for
power consumption, robustness and reducing costs. Trends in technology
mean that the components we use will continue to become more efficient
and cheaper.

All of our technology is open source so you're welcome to use any of
it or contribute to the project in any way as well. If you'd like to
get involved a good way to get started is to start a conversation on
our mailing list/forum at
http://groups.nzoss.org.nz/groups/projectcacophony. We're interested to hear about your project and the way you've tackled the
challenges you've encountered.


Grantryan Oct 8, 2017

Thanks for your feedback. Just following up from Mennos comments below the reason that we have gone for higher resolution heat camera is that we were finding a lot of animals walking straight past the traps or tracking tunnels - see the end of the video in this blog post https://cacophony.org.nz/issues-and-opportuni...e-capture-traps. The higher resolution and power consumption we think is worth it so that we can see a full picture of what is going on and these technologies are getting cheaper and using less power quite rapidly.

As with such a big problem of predator free there are lots of good ways to tackle the problem and completely understand your project is more likely to be useful in the short term. Happy to put you in touch with the folks at ZIP that have also been playing with lower resolution heat cameras.

With our project we are taking a purposefully different approach that is more long term of trying to make a device that can detect everything over a large area - we will then be able to lure, auto detect and eliminate everything eventually. This may seem overly ambitious but we think making the ultimate device then making it cheaper is a good complimentary way to tackle the problem. More about the logic of this approach here https://cacophony.org.nz/moore%E2%80%99s-law-...tor-eradication. But as I said early we need to try all approaches.


View all replies (2)

Hannah Smith Oct 4, 2017

I'm so enthused about this project. The fact that as predator populations get smaller food sources for the ones left increase so that baited control systems no longer work is such an important concept. This is the most important project in this competition in my mind. On a lighter note how about audio or pheromones implanted in the cats eyes down the middle of all our highways :)


Menno Smits Oct 4, 2017

Thanks for your support! Your "cats eyes as predator lures" idea is very creative :)


Jennifer Mcguire Oct 9, 2017

Love the cats eyes idea Hannah!


Hannah Smith Oct 9, 2017

I know! I'm very proud of it! :)


View all replies (3)

Hannah Smith Oct 4, 2017

PS I can't figure out how to "follow" cacophony.


Menno Smits Oct 4, 2017

There should be a Subscribe on this page on the right hand side, below where you vote.

More generally, you can follow us on Twitter for project updates and relevant news. We're @birdsongnz.


Brent Beaven Oct 6, 2017

Great engagement on this idea. As I seem to be saying for everyone, lets get more specific.
As well as describing the system of sound lures, cameras and AI, I think you should pick the element that you are looking to progress through these awards. My thinking is that the sound lures are a project in themselves. Perhaps it is the camera recognition work at this stage? Certainly that would seem to have the potential to link to a range of future applications....island biosecurity systems spring to mind.
I will check on progress mid-week.
Good luck,


Grantryan Oct 8, 2017

Hi Brent - thanks for your comment. We had suggested the camera movement tracking as a specific project that this award could help accelerate but open to other suggestions.
The reason we chose this is we have the project mapped out to update the camera so it can identify using AI all the predators. We need to complete this before we can more systematically test the sound lures.
A summary of our initial sound lure testing can be found here https://cacophony.org.nz/possum-reality-tv-di...-testing-update and there are links to many other projects and papers here. https://tree.taiga.io/project/the-cacophony-p...n-status=268961. By having a device that can automatically detect the predators these tests will be able to be done 1000s of times more efficiently.

The other non-obvious problem that we have noticed from testing is that when we get a positive result from a sound lure and the predator is not eliminated we have effectively trained a predator to ignore what could be a great lure. Current predator traps work as little as 1% of the time https://cacophony.org.nz/existing-predator-tr...k-little-1-time. We have been testing live capture traps that are about 10 times more effective but that still has issues

This is the reason we put the predator elimination higher up the priority list - it will make the digital lure testing more useful.

Happy to hear another suggestions on how to test the sound lures more effectively.


Brent Beaven Oct 12, 2017

Hi. That sounds like a good approach. Focussing on the camera and AI gives this entry a real focus. Good luck with the stages and judging.


View all replies (2)

Jennifer Mcguire Oct 9, 2017

Great project with valuable application. Very detailed and so much great work gone into development already. Outstanding.


Joanne Jackson Oct 12, 2017

Interesting concept. These devices have high power consumption and solar efficiency would be compromised under a canopy - can you resolve this?


Susanna Finlay Oct 12, 2017

Thanks for your question.

While we acknowledge that power management is going to be a challenge, it's not something we're too concerned with at this phase of the project. Right now we're focused on proving the concept works.

Once we have demonstrated our approach is viable we will focus on issues such as power sourcing and consumption. It's likely that we will work to have custom hardware built which suit our power and robustness requirements. Additionally, we are confident that the kinds of components we need will continue to become more efficient (and cheaper) over time.


Joanne Jackson Oct 12, 2017

Thanks for your reply Susanna. Do you know how you will get web connectivity in remote areas?


Menno Smits Oct 12, 2017

Although connectivity is important during our development phase where we are rapidly iterating software and data, the final devices should be able to operate with some autonomy and won't necessarily need continuous connectivity. It is anticipated that our devices won't have to stay in one location permanently but will be moved through an area so any data transfer required could happen at this time.

We will also consider the use of mesh networks or long range (albeit low bandwidth) network technologies such as LoRa to maintain connection with devices.


Grantryan Oct 12, 2017

Just expanding on the other comment related to power and data connectivity. Both of these things are on trends that have seen dramatic improvements and are projected to continue these trends. On the scale of making NZ predator free by 2050 we can assume the cost of power and data are going to be extremely unlikely to be limiting factors.

The not obvious point mentioned by Menno is that when we achieve 100% luring, detecting and elimination then the devices do not need to stay in one place. They can be moved in lines to eliminate with only a few left for low level detection. This again means that power and data and not such large issues.


View all replies (4)

Michele Frank Oct 13, 2017

The idea has been progressed to the next milestone


Squawk Squad Oct 14, 2017

Outstanding work The Cacophony Project!