It's neat to see this on hacker news. Good to hear that folk are noticing that our system works! I've been working on embedded systems at Zipline for 6 years, and oddly haven't made it out to Rwanda yet. Everyone there seems to know my name, though, because of all the weird problems I've helped remotely debug over the years. I think Zipline's flight operators deserve the lion's share of credit for our success in Rwanda.
I was stationed in Rwanda for several years, and visited your operations centers several times. Truly awesome stuff, that's also helped Rwanda in a bigger way, by giving the country well-deserved recognition for its propensity towards innovation.
Any change we can get it contact with you? For our project www.4x4electric.com we would like to visit different Zipline projects in Africa to generate more attention for this kind of projects
Cool! I raced solar cars in university, and now I volunteer with the American race organization. Of course, the University of Delft was my team's nemesis! :-) I'll email you from your website contact info.
I'm curious what you do for the American race organization - I'm assuming you are referring to American Solar Challenge. I've dabbled with solar cars, too, and it is rare to find a fellow racer.
Yep, I am one of the electrical scrutineers for ASC. I help inspect all the teams' cars for compliance with the race regulations as well as for general soundness to make sure there aren't any safety concerns.
I've watched a (the?) video that explains how you developed the landing/catch mechanism for the drones a couple times when it's gotten posted here over the years, and it's seriously one of the coolest things I've ever seen, both the iterations and the final design you arrived at.
I think the Zipline drone systems would be great for remote sensing of environmental data; for example of destruction of the Amazon or bleaching of the Great Barrier Reef. Have you looked into those sorts of markets?
I imagine there's already some reasonably cost effective ways to do that. For example, existing satellites or piloted single engine aircraft like Google uses for their mapping.
Zipline's aircraft tend to fly the same trajectories over and over again, whatever's lowest energy, fastest, or avoids flying over populated areas unnecessarily. The vehicle isn't really designed to fly repeated grid patterns. We're also hesitant to intentionally collect aerial photography. It isn't necessary for our core business of efficiently delivering medical supplies, and could negatively impact the perceived utility of our system.
You should check out Saildrone. They make autonomous boats for collecting ocean science.
https://www.saildrone.com/
My job specifically? Experience with microcontroller programming, as well as software development in general. Familiarity with linux. It helps to have a decent understanding of physics and mathematics. I myself studied computer science in school, and I think that foundation helps me to write more elegant solutions to certain problems than I would know how to do otherwise.
I seem to be pretty good at drilling down into complex problems and finding relatively simple solutions. I am also pretty good at pushing back on things when they don't pass my smell test, and am willing to crack eggs as needed. Those are skills I practiced quite a bit in my solar car racing days. It's the sort of thing that many engineers struggle with, and are often afraid to do. In a good organization, though, it's the sort of behavior that others respect and appreciate. It helps to have an optimistic, easy going attitude.
Look at our career site, https://flyzipline.com/careers/ (we are hiring!) Even if you do not see a perfect match, consider applying -- most teams value good engineering skills higher than a perfect match of some technology lists.
Cool! People working in embedded tend to have a wide variety of backgrounds. Detailed knowledge of various internet protocols would be quite relevant and valuable. Have you worked in any compiled languages much?
I’ve done some C with Arduinos before and a bit of rust. You wouldn’t happen to have any internship type of opportunities, would you? or maybe some one-off tasks that I can be asked to solve, given some basic direction
In the United States and the United Kingdom, 80 percent of the population clusters around urban hubs with high-traffic hospitals and blood banks. In African nations like Libya, Djibouti, and Gabon, about 80 to 90 percent of the populations live in cities, too. But in Rwanda, that number flips: 83 percent of Rwandans live in rural areas.
DON’T BE FOOLED by Rwanda’s rural demographics; the country has a reputation for leaning into health tech innovations. Rwanda’s universal health care system reaches over 90 percent of the population. In 2009, the government piloted a phone-based program, called RapidSMS, to track and reduce maternal and child mortality. By 2013, RapidSMS connected 15,000 villages to the country’s wider network of doctors, hospitals, and ambulances.
The cause and effect might be flipped there. High traffic hospitals are built close to urban centers.
As cynical as this may sound that 90% figure is v optimistic, virtual access to doctors and hospitals is not the same as physical access. But this is massive progress nonetheless.
There's this new podcast What's your problem? from Pushkin Industries, about the big problems entrepreneurs face and try to tackle, and one of their first episodes was about Zipline:
The big challenge they talk about is bringing the service (which already works in Rwanda) to the US, where aviation regulation is very different.
(Cautionary Tales with Tim Harford, about "awful human error, tragic catastrophes, daring heists and hilarious fiascos" and what one can learn from them is also entertaining and informative.)
I live in a region of the United States that is significantly less populated than Rwanda (larger, a few hundred thousand residents) and don't see what the opportunity would be in health care.
I guess you could discard less perishable products if they were moved around for use.
Why not say why if you think there is something interesting over there? Don't assume I care to listen to 25 minutes of whatever to maybe figure out what you think is interesting about a podcast.
Listening to a couple minutes, it sounds like they talk about expired medicine. Great. In any case, I don't see that there is really much of a market for moving 4 pounds at 60 miles an hour in the US. We have decent roads everywhere and the resources to have extra things where they might be needed.
A concrete example. Like I say, I live in a region that is larger than Rwanda, with something like 1/75 of the population density. There is a hospital with an emergency room in town here, the regional hospital is about 55 miles away in a straight line, a 75 minute drive if you don't speed, and then a real city is about 100 miles away in a straight line, a 2 hour drive if you don't speed (and it's similar for most of population in the region, living close to an emergency hospital and not all that far from a larger hospital).
What logistics are you going to institute where frequent small deliveries make a difference compared to regular vehicle based deliveries?
I guess if it is cheap it can be used for home deliveries.
I watched the videos about this, and compared against the videos of GoogleX Wing. I just don't get how they can be so different, or more precisely, how or why Wing is doing what it's doing.
This Zipline stuff is launching simple, low-cost vehicles, dropping very valuable (even life-saving) things from parachutes as simply as they can.
Google is launching very complex (part helicopter, part fixed wing) aircraft, lowering $5 coffee drinks to people so it doesn't spill.
How can it be so different? Do they just... have too much money and are over engineering something because they can? or have lost the plot? Or will they eventually get into delivery of some good that will justify the cost?
Or maybe I'm being a little harsh and it's just a reflection that in developed countries, life-saving needs are already taken care of quite well at the local level and all we have are coffee drinks to satisfy people's urgent desires nowadays.
Yes, "they just... have too much money and are over engineering something because they can". But also they are trying to fly precisely through extremely busy and regulated airspace to make a pin-point delivery of a low margin product. So you really do need a hover-capable vehicle and a method of landing within ~1m of your target. This may or may not actually be a good business model, but as you point out, in fancy countries we already have roads and plentiful hospitals.
Zipline's drop zones are larger and their operating area is so rural (not even roads! let alone air traffic!) that they had the freedom to choose simple, reliable design strategies.
Btw, how did you feel about being part of Google and then spun off as a stand-alone company (if you were during that phase)? Did that cause big changes in company strategy, or more practically, people working there to regard it differently or want to leave? Do you think the outlook is good for the company?
Where the development team is based physically is the main factor I think. You can deliver the hot coffee to the manager who makes the decision wether the project is worth doing at scale or not. They might have some doubt, but there will be an actual coffee in their hands.
Convincing the same manager to green light a potentially life saving project for stuff they'll never see with their eyes, on the other side of the earth is way more difficult, that's just human nature I think.
This is also why I think empowering (~= sponsoring/supporting) local people to solve themselves their problems is critical and can't be replaced by drop in solutions sent from big brains who have no skin in the game. It sometimes work, but it would be the exception and not the rule.
This is so important. Just using the name of the place will get people to Google it and find out more about their people and culture, and pragmatically, learn about their solutions to problems and problems that may need external help solving.
I remember hearing about this in a video on drone delivery in Phoenix AZ, which due to weather conditions seems like a great place to fly drones. The big takeaway was drones make great sense in places where no delivery infrastructure by ground exists and the airspace isn't highly restricted, and that they don't make tons of sense elsewhere. Also that autonomous takeoff and landing are hard to get right, and plenty of people in cities rent small apartments and don't have a space like a yard or a large balcony big enough to just have a drone easily, safely automatically land.
I saw another video about a network of automated delivery bots on a very famous and chaotic Minecraft server, 2b2t, run by an opportunistic software developer. Players who want large amounts of valuable in-game items make a small payment of real life currency to this player† and request it to drop them off at a certain set of in-game co-ordinates. The bot player uses a glider to get there and delivers large payloads of valuable items taken from a secret cache. Then the bot self-destructs so that they respawn at the secret cache instead of potentially revealing its location by going back. A very large chunk of the server's player base of players on this server have no home base, and just roam a very chaotic landscape which has mainly just large main highways, which get griefed often. It's only a game, but I think that models to us how relatively wealthy nomads living in very rural area/the wilderness would be a good target market for drone delivery.
I think for automated delivery systems in well-connected/developed areas, we want a system that uses small rails/tubes like what we see in Brave New World.
† This is against Minecraft's terms of service, but they do it anyways.
- The drone is not a quadcopter, they are too inefficient. Mini plane design.
- GPS device is directly attached to the swappable battery unit to ensure it is always connected to a satellite, eliminating satellite connection delays when the drone powers up.
- It is launched into a sky by a powered rail. It doesn't need to land to deliver the goods, the items get parachuted down. No infrastructure needed on the receiving end, the item just drops from the sky.
- For landing the drone flies past an aircraft arresting wire while hooking onto it which decelerates it and causes it to swing down (suspended above the ground). Easy to retrieve.
It's been great for streamlining operations. The first generation of aircraft we deployed to Rwanda didn't yet have blind-mate connections for the battery pack. Operators had to reach in and connect two connectors. The plastic tabs that retained the connectors would tend to break off after a while, requiring the operators to secure them with a zip tie, on every flight! I forget what generation of vehicle was featured in Real Engineering, but I suspect we're one generation past that one, and the batteries are a bit more streamlined from that.
I haven't paid attention to the statistics in a while, but these days the busiest distribution centers can launch a vehicle once per minute or so, and they do.
Yeah, I just skimmed the video, and that aircraft is one generation behind the current design. It was a good aircraft! The new one looks very similar externally, just a lot more optimized internally.
One of the sound bits I heard while skimming was that the recovery system misses 10% of the time. We've dialed it in to be considerably better since then. I'm going to forward that timestamp to a particular coworker and see if it causes his eye to twitch... :-)
Man that's such a cool job/place to work. I was curious, I don't know if you can answer this or if it's public info.
Do the aircraft boards/computer get tied to some kind of simulator and (autonomously) fly around/run the actual control surfaces (in reality) as it flies in the simulation that would be neat. Although it might not make sense.
Yep, that's a thing. It's called "hardware in the loop". There are indeed control surface servo motors plugged in, and they move around to wherever the flight computer commands them to based on what the simulated vehicle is doing. Unlike a real vehicle, though, the boards and wire harnesses are quite a bit messier. All the important wires run through additional equipment so that folk can automate tests where things break.
One thing that we don't have plugged in normally is the propulsion motors. The noise would be rather obnoxious considering the HIL setups are in the office. Also,
I believe we still omit the wing tip LEDs; not only would they be obnoxiously bright, but they would melt themselves for lack of actual airflow!
Its not just the swappable part. I think they have modularized all the electronics and GPS communication systems inside it, so that it is always booted and on standby. As soon as it is incorporated into a drone, it will be ready for work.
Another innovation is the clip on wings.
I think this is clever engineering and a necessity for their time to fly requirements.
Also, Wendover have a fantastic Youtube channel that’s focused on global logistics. It’s endlessly fascinating and extremely well executed. Strongly recommended.
It was really interesting to watch. General idea discusses the problems it faced in the US in seemingly optimal circumstances (airspace restrictions, density, existing delivery networks) and how it does live and help out in other places, highlighting zipline as an example.
At the same time you can't fly drones pretty much anywhere in CA, and the situation is even worse in TX. In TX you can't even make photo/video from drone and it is a crime of trafficking in such photo/video ("land of the free" despising government overreach :).
Drones is a transformative technology, and we as usually go for 19th century Red Flag car laws.
That's a loaded term. It means whatever you want it to mean. If you are a drone enthusiast, it means you fly wherever you want, take pictures of whatever you want. To other people, it means freedom from having a drone buzz over their house taking pictures.
I read your post. I go, surely with what he’s stated if it’s true, it wouldn’t be hard to find.
I put in my iPhone search engine “Texas drones animal cruelty”
And, in the first results, are these receipts you ask for.
I am begging you to simply use the search engine that takes no more than five seconds of your time rather than posting asking for “receipts” or “sources” & “citations” …
It seems like a fair question, though. I did your google search. All of the hits that immediately came up were from sources that clearly have an axe to grind.
In any case, it looks like the law got mostly overturned a couple weeks ago, if I'm reading correctly. I imagine there will be ongoing litigation. Even without "ag gag" legislation, there is no shortage of people who really don't want drones flying around their neighborhoods taking pictures.
Thanks for telling me what not to post here. Can you be more explicit as to what I am allowed to say? You appear to be more helpful than the HN guidelines, which aren’t as thorough as you are.
P.S. my Web searches didn’t come up with reliable-looking results, which is why I asked the illicit question.
Can you tell me what about your web searches didn’t align with what the parent said?
All I’m trying to say is not to post the most low hanging of fruit. Something to the tune of “I’ve looked & you appear to be incorrect” is much better than “receipts?”
Great article - this is the first time I am reading about the positive effects of drones where it has actually saved lives. I have also read that Rwanda is doing very well recently since their terrible civil war in the 90s.
Well they’re just miniature aeroplanes, so any resources on aeroplane design should get you started — there’s probably more info out there on designing aeroplanes than quadcopters.
Can I just say I am jealous that Rwanda has universal health care? I don't know all the details, and I do think it is easier to govern smaller countries, but dang, that's cool.
My inner seceng wonders ( which likely cannot be answered in an open forum :) ) about the security of the system since the payloads are likely valuable - meds and all.
