I learnt about Zipline in a talk at JuliaCon by Tucker McClure who spoke about how they use Julia for designing and tuning their aircraft. Had the great fortune of spending an evening at the conference to learn more about the company itself, how it got started and their use of Julia.
Zipline is such an awesome company. I love that their largest target market is Africa. Almost every high-tech startup starts with conquering the US en EU markets even though double-digit growth of economies is now mostly happening in emergent markets. It’s one of many interesting points touched upon in Hans Roslings last book (Factfulness - a must read!) and here we have a company that did exactly that. And they are doing it with such cool technology as well!
They are doing high value deliveries in low infrastructure environments.
I live in a rural region in the US and can pretty much drive anywhere within 50 miles in about an hour, with hundreds of pounds of whatever. There's no market for slightly faster deliveries of 3 pounds.
> Almost every high-tech startup starts with conquering the US en EU markets even though double-digit growth of economies is now mostly happening in emergent markets.
I would inject a healthy dose of scepticism here.
To the chagrin of American HN readers, I'd say that while the total digits are incomparable, the number of technical enterprises is certainly bigger outside of USA. Most of those companies can't find a single reason why they have to be in USA.
USA technology sector is overcrowded, and overfunded, and I am not the only one person around who can make this observation.
> Zipline is an American medical product delivery company headquartered in South San Francisco, California that designs, builds, and operates drone aircraft.
The website includes a video [0] that explains how it works in more detail too.
It's quite interesting how they evade issues with landing and lifting the plane, lifting the plane by shooting it off a rail seems like a common approach but landing it by catching it using a small hook at the tail of the plane isn't that common I presume.
On their website it says "Gas combustion vehicles break down, get stuck in traffic jams that prevent urgent response, and put human drivers at risk behind the wheel, particularly when the route is rough and treacherous. Zipline’s drones are battery powered and fly quickly and directly to their destinations, leaving ground vehicles behind."
That sounds like an unfair comparison, a more accurate comparison would have been to gas powered drones. Is it still the case for gas powered drones that they're more unreliable than electrical drones? It's probably because of the weight of a gas powered engine in such a small drone no?
Ok, the first 4 were lost but this is a huge journey with limited fuel and bad weather. And navigation was a problem too.
The big benefit of gas powered in larger vehicles is both the easy fast refueling and the higher energy density of fuel. Combustion engines for model airplanes are heavier than their electric counterparts because you can only scale things down so much, you still need to contain a constant fire and the associated pressure.
And in this usecase the slow recharge isn't really an issue anyway. It also makes sense for marketing reasons, being 'green'. Even though if fuel were used, you could probably run this entire company for a year on the fuel a semi truck would burn to drive 100km :) I don't think it would be a bad thing for the environment if a company like this would use fossil fuel, it's still a lot less wasteful than the alternatives.
I wonder if a pulse jet (as used on the V-1) would be an option?
I've seen hobbyist level plans for them, and a pulsejet requires zero moving parts, so reliability should be excellent. The geometry of the engine does all the work.
I doubt slow recharge is even an issue since they can just pop in a new charged battery pack and put the depleted one on a charger. With swappable battery packs a drone can be relaunched within minutes.
> landing it by catching it using a small hook at the tail of the plane isn't that common I presume.
A lot of military drones have a hook at the end of one wing, and they hang a vertical wire under tension from a boom arm, with a differential gps receiver at the top of the wire. Ends up it's all accurate enough for the drone to catch the wire. I presume zipline went with their more complex approach as wings strong enough to just snag and stop would be past their weight limits.
> That sounds like an unfair comparison, a more accurate comparison would have been to gas powered drones.
It's a comparison for customers, showcasing the difference between zipline's solution (electric drones) and the existing solution (gas trucks) to the problem zipline is trying to solve (medical logistics). If gas drones or electric trucks were to enter the space then it should be updated.
What’s often overlooked about Zipline is that they started as Romotive, a company that made iPhone—powered robots for kids.
Through the sheer force of willpower by the CEO (Keller) they found a new path and have become what they are. It’s a great startup story of how to persevere.
Disclosure: My firm was one of their first investors.
Hi there! Ryan from the Zipline founding team / Romo product lead.
We always had aspirations of building robots that solved real problems. Our early plan was to slowly make the toy robot more and more useful. But suddenly in 2013, it felt like the robotics industry was moving much faster than we expected. We threw out the plan and started searching for the highest impact work we could do.
We spent months talking to customers, exploring technical approaches, and decided to race to bring something to market. Within two years, we were making daily lifesaving deliveries by drone.
It's wonderful. The word "Droneports" is becoming mainstream.
Can someone shed light on their economies of scale? What are the challenges they may be facing?
It is also interesting to know the technical and engineering challenges they may be facing with the technology. Are they drones completely autonomous? Command Control can change the route mid-flight? How do they do communication in long-range? How do they handle faults in-flight?
> Can someone shed light on their economies of scale? What are the challenges they may be facing?
A drone like zipline's latest model cost at most $500 in low volume manufacturing.
Making at least 10000 of them in one go will likely to halve the price, and I see them going there over last few models.
With 100km range, you have to put extra droneports on the triangular grid.
To link the US West Coast North to South you will need a minimum of 20 droneports. Add 10 more of them reach most major coastal cities.
So, making a commitment to buy 10k drones over 2-3 years, and operating 20 droneports should cost around $4m-$5m a year. With just $1m going to drones themselves.
Making $5m a year from small package express delivery on Seattle-LA route should be not that hard. 200 packages a day at DHL prices.
You can halve the droneport costs if you make bigger drones for longer range transport in between two major hubs, but that's the problem.
100km is a practical limit of an electric fixed wing drones that carry sub 1kg payloads. If you want more than that, you either make them really big, or venture into play with supermaterials like CFRP, or IC engines/turbines. The cheapest PT6A variant costs $500k...
From the video referenced in other comments, the parachute is disposable (looks like it's made out of a bit of paper or plastic and packing tape), and the range is 160km round trip, the service area is an 80km radius around the droneport.
This stuff is awesome, and I really hope we reshape society for automated last-mile.
Trucks and human drivers walking up to doorsteps are insanely wasteful, slow, and expensive. I wish more apartment buildings and metro neighborhoods were into pneumatic tube systems, too.
The last time they were tried, microcontrollers and fiber optic networking weren’t as good as they are now. A packet switched automated last mile system for everyone would be incredible.
I suspect you’re getting downvoted for the pneumatic tube sentence, because tube systems like that have proven to be way too inflexible for a rapidly changing city. And expensive. And hard to maintain.
I haven't really seen any research on packet-switched pneumatic tube networks, or any kind of more complicated routing. Every time I've seen one in person it has been more or less point to point, or point to mailroom. I think modern sensing tools, either NFC or high-speed cameras, have the potential to enable some significantly better topologies than what I've seen in real world pneumatic networks.
Also looking at projects like the hyperloop, I'm wondering if a smaller version linking cities would make sense. We already have infrastructure for transporting oil and natural gas through very long pipelines. How about actual packages? If the hyperloop can make sense, I feel like inter-city mini vacuum train probably can as well, and linking two cities together doesn't require any great flexibility.
I've a friend who thinks drones are ultimately a better solution than tubes, which is also quite plausible. I don't really care much how it gets done, I want <10 minute automated delivery directly into my residence (hot food stays hot, cold drinks stay cold), and I imagine millions of others do too, if for no other reason than to make homes entirely without kitchens or refrigeration feasible.
So what prevents the package from being dropped on someone unaware of the delivery? How heavy and how fast do they fall? Their website doesn't seem to cover this, the mention of a predetermined drop point makes me think that it is just gps. And the box in the video seems to drop pretty quick.
The target drop zone is about the size of a parking spot, and you'll hear the vehicle coming before it drops the package. They're fun to try and catch before they hit the ground...
Drones are loud and pretty energy inefficient compared to hydrogen balloons. I understand large balloons like Zeplins cost a lot of lives, why can’t we make mini Zeplins for <1kg payloads. Or <5kg payloads. That should scale right?
A drone could be attached to the balloon and it would help reduce the energy needed to fight gravity and could probably cruise for a much longer distance.
Square cube law says no it doesn't scale well I am afraid. The weight of the skin scales on surface area and the lift scales on volume of the displaced air.
You can tweak for thicknesses and pressures some - but hydrogen is also corrosive in addition to being flammable and explosive.
That would be pretty cool. I question how practical it would be in real world weather conditions, though. Also, with that much surface area to mass, I can't imagine it could go very fast efficiently. A big benefit of Zipline's vehicles is the capability for emergency just in time deliveries.
Minimizing footprint was our #1 priority to help our customers who have considerable data limitations. Also getting most everything to load nicely with the first resource was key for slow connections (10s of seconds per resource).
This has the nice effect in the US of being pretty much instant
We changed the title from "Zipline - Vital, On-Demand Delivery for the World" which is too marketspeak. If someone suggests a better (more accurate and neutral) title, we can change it again.
Recommend the 2019 JuliaCon video on how Zipline uses Julia for vehicle simulation: https://www.youtube.com/watch?v=K8rPZVotroY
And this video by Real Engineering that does a great job of explaining how the whole thing works: https://www.youtube.com/watch?v=jEbRVNxL44c