Hey everyone,
I've now been working on this project for 10 months. I have great news to share, and I think now's a good time to share our progress and where we're headed next.
I've started this project based on some very bold assumptions. The great news is that they were mostly on point, and we're now a lot more confident in our original hypothesis : that we can make a cost-effective dent to malaria by building cheap mosquito-hunting drones.
The first insight came from studying mosquito life cycles, and led us to conclude that we wouldn't need nearly as many passes per hour as we initially believed. I initially believed that, to protect a fixed zone from mosquitoes, we'd have to pass at a very high frequency (~1 time a minute) everywhere in the zone, which would be prohibitively expensive at scale. We've now realized that we'd only have to do this on the border of the zone we're defending. To protect a 1km² zone, we can make high-frequency passes on the border of the zone, to avoid mosquitoes coming in and lower-frequency passes inside. Mosquitoes take multiple weeks to reproduce, and females need two long (20min) nectar-seeking or blood-seeking flights to produce and ley eggs. By doing a pass every 20 minutes, we can divide the probability of each successful reproduction by 85%. That would, over the weeks, allow us to eradicate mosquitoes inside the zone.
Our second big success has been on detecting mosquitoes. We've been able to shrink the size and weight of our sonar sensor. Our latest mosquito detector works at three meters, weighs a few dozen grams and should cost $20 at scale. I had initially budgeted for three times more cost, size and weight.
This solves a lot of other issues. One of our concerns was on the noise made by the drone. With a lighter sensor, we can use a lighter drone, which makes a lot less noise and at a higher, more attenuated, pitch. We aim to go as small as a tinywhoop, which would also solve most issues with regulation. They're too light to hurt anything and are extremely quiet. The only issue with them is battery size. We don't expect to have more than 10 minutes of flight time per drone, but they're cheap enough to not be a problem. We'll also investigate doing battery swapping to increase capital efficiency.
Our work on mosquito detection is now complete, and we have a state of the art but obsolete for us ultrasonic phased array sonar. We've decided to open-source it, as it can be used for much more than mosquito detection : air leak detection, people counting, autonomous robots, ... you name it. Anything that needs an accurate 3D map of an environment in all lighting conditions.
We've also done tremendous progress on simulating our system. We now have a complete drone simulation, and are prototyping algorithms for collision avoidance and mosquito hunting. We have great results in simulation so far.
On the hardware side, we have sourced all of the off-the-shelf components we need for our first mosquito killing prototype. We still have to write a lot of embedded code and integrate a few mechanical parts, but it's taking shape. I expect us to have a drone flying in the air by January, and it catching its first mosquitoes in the coming months.
As to the people, I've gone back to school in September for my last year and have recruited Clovis as a co-founder. We're almost working full-time on it, in parallel of our light curriculum. We've done market research and user interviews, and estimate that we could fund our long-term development costs by selling mosquito-hunting drones to individuals throughout the US. I promise that the people cited explicitly or implicitly in my last paragraph will be the first to be delivered!
In terms of funding, we've got great profiles graduating at the same time as us, and we'd like to hire them. Therefore, we're planning to do a fundraising round by April of next year. We have great feedback from angels and VCs, and are working on the pitch for grants from nonprofits. It's mostly a matter of finding the best path for us personally. This post is only a small update, and I plan to write a detailed, budgeted, public white paper on how we could eradicate malaria with mosquito-hunting drones. Our current numbers show that we could eradicate malaria for a very realistic amount, but I won't publish more details before getting our assumptions checked by more people.
As a final note, I'd like to thank the people who have supported us along this journey.
Firstly, Scott Alexander and ACX Grants for giving me the financial support to work on it full-time and for introducing me to the kind community of ACX readers, many of which helped me directly in one way or another. I'd like to mention Austin Chen and Rachel Weinberg, which have helped a lot by handling the technicalities of accepting donations. I'm also deeply thankful of Austin's precious advice. I'd also like to thank Jan Steckel and the Cosys-Lab team in Antwerp for giving us lab space and precious technical advice for six months. Finally, I'd like to thank all of our independent donors. And, a big thank you to the many others who have helped us get this far - you know who you are and your support means a lot!
Useful links
My twitter for more frequent updates.
The public page for our latest Sonar sensor, which shows most of its capabilities.