Iterating with Hardware: Challenges in Building a Drone Delivery Product

Alan Hicks at UXDX EMEA. Video: https://www.youtube.com/watch?v=OYDNftJ6fyg

Readable transcript: edited from the recording's captions for readability (fillers and false starts removed, punctuation and section headings added). Wording is the speaker's own. Timestamps are positions in the video. Names marked [?] could not be verified against the audio.

Why Manna exists

[00:00:00] Hi everybody, Alan here. I'm here to talk to you today about iterating with hardware, the Manna drone delivery story, and just to let you know what we're doing, what we've done, and what we hope to do in the future. Our mission at Manna is to improve the world by making lightning fast suburban deliveries affordable, green and safe, and we do that by essentially magical three minute deliveries from the sky, using our drone technology platform.

[00:00:28] A lot of people ask us why we got into the business, and the why is very simple. Current last mile delivery is just very, very unsustainable and very, very expensive. It costs up to ten dollars per delivery, and that's mainly down to labor costs and running costs and just pure operational costs of road-based delivery. It also suffers from quality problems. To make the unit economics of last mile delivery even somewhat attractive for road-based delivery, orders have to be aggregated together. So it's very, very common for a delivery driver to have to take five or even 12 deliveries at a time from a restaurant, for example, and deliver those to the houses, and that leads to very erratic quality.

[00:01:17] What can happen is, it's great if you're the last order into the car and the first order out, fantastic, you've got nice fresh, hot food. But if you're at the other end of that scale, the quality of the food is going to suffer. And the third reason that we're doing this is the negative societal impacts. Current road-based delivery is very, very unsustainable for the environment, it causes a lot of traffic on the roads, and it's just not a very efficient way of getting small goods around for delivery.

[00:01:50] So with all of those reasons in mind, Manna was born, and we were really born to transform last mile delivery. We can do that because we can operate at a very, very low cost, mainly through automation, and really, really high quality products. We're delivering one to one, one product at a time. We can get you that product there in roughly two minutes 40 seconds, so we really have that highly consistent quality. And again, a positive societal impact: zero emissions, fully electric, and the vehicles are safe and autonomous, and you don't even hear them, they're 50, 60 meters in the sky, removing traffic from the roads. So all around it's a positive experience for everybody.

How a delivery actually works

[00:02:34] Just to tell you a little bit about the process and how it actually works. Everybody's used to their traditional aggregator apps, so the customer goes online onto their app, they place their order as normal, and the vendors and merchants prepare the order as normal. The only difference is, instead of the food or the payload ever being loaded into traditional road-based delivery, it's loaded into an aircraft, and that aircraft will then autonomously fly to your location, taking no stops on the way, and does that at about 50, 60 meters.

[00:03:11] When it gets to the delivery zone it drops down to about 15 meters, the doors open, and the payload slowly drops to the ground, and then the craft returns home for another delivery. So very fast, very efficient, and again, very clean and safe.

[00:03:31] Just to tell you a little bit about our platform. To enable the business we've had to build a drone and a bit of physical hardware, and the reason we had to do that was that consumer grade off-the-shelf drones just aren't sufficient in terms of levels of reliability. What we've done is we engineered an aircraft from the ground up to be highly reliable and to be basically a workhorse. We fly our aircraft up to 12 hours a day doing constant deliveries, and to do that safely requires a lot of engineering effort, and a lot of time has gone into both the hardware and the software to enable that system to be highly reliable.

[00:04:15] As you can see here, there's a current view of the aircraft, and it's got multiple motors on each arm, and that means that we've got redundancy on each arm. Internally in the craft we've got multiple batteries, and then we also have a triple redundant flight control system, and that's really at the core of what makes our technology very, very safe to operate. We've got three independent suites of flight sensors and they're constantly voting between each other to ensure that the aircraft is making the right decisions and flying safely. The body of the aircraft is designed for lift, and to allow us to deliver the payload quickly, safely and efficiently.

[00:05:05] Our typical cruise speed is about 85 kilometers an hour, and we fly as the crow flies. I'll take you through the journey as to how we got there in a few slides. We're 90% cheaper versus car delivery, and that's mainly down to automation. We can automate a lot of what happens in the delivery process, which again increases the consistent quality of what we can do. And we're designed in a certified environment. Again, we'll talk about that a little bit later on.

Moneygall: where it all began

[00:05:41] So, to my journey. We've been doing this for about three, three and a half years, and it's a really, really interesting space to get into. It's very much the beginning of an industry as we see it, and there's very few companies in the world that are actually operating a drone delivery business, and certainly very few that are doing it successfully. We're based here in Ireland, which again is a great place to be doing this. We've got lovely weather for aircraft testing, which kind of gives us a little bit of an advantage.

[00:06:15] As you can see here, we've got some different dots here, so this is just to explain the story that we have now. We actually started our journey in Moneygall, and I'll talk about that in a second. We then went on to Oranmore in Galway, and we're currently live in operations in Dublin, in Balbriggan, just near Dublin Airport. We also, the orange dot here, we have a test facility down in Offaly where we're flying about 600 flights a day, and we manufacture down there and we do all of our commissioning, safety testing, parachute testing and everything, down on site in Offaly.

[00:06:53] So, Moneygall, I suppose, where it all began for us. It was an interesting time, it was back in 2020 when the world was going through some changes. Actually, we as a company were based in the Innovation Centre in UCD in Dublin, and our initial trial was actually to go from one side of UCD to the other. So reasonably short point to point, A to B trials, relatively simple. And again, everything in aviation is crawl, walk, run, so you start simple and you build up confidence on all levels.

[00:07:26] That was what we intended to do. We intended to go live in March 2020, and then obviously the world changed, and the campus in UCD closed down, so there was no students to deliver to, and we had to put that operation on pause. But a few weeks later we actually got a call from the HSE to see if they could use our technology to help individuals cocooning in different parts of the country. We jumped at the chance, it was a great opportunity for us, and we actually ended up going to Moneygall, and we based our drones at the Barack Obama Plaza.

[00:08:07] What you can see here on the right is orange lines, and each orange line represents a delivery flight. These are real deliveries, customers are ordering real things, us delivering prescription medication to people cocooning. A really, really interesting journey for us. We had gone from doing a point to point, an A to B operation, to, a few weeks later, going into actually people's gardens and delivering into gardens. That provided a lot of technical challenges: how exactly do we know where the pin is, are we dropping in the right location? And a lot of operational challenges. At that stage we didn't even have a website or a mobile app, we were actually taking orders by phone. But it worked, and we made it work. There were a thousand people there that were very, very happy with the service, about 300 homes.

[00:08:58] What you can see here is our base here at the center of the orange, and the routes kind of follow what we call highways, and you can see that they're going over mostly green areas. This is down to the safety cases. Obviously we're flying an aircraft 50 meters in the sky and so safety is a permanent concern, and again we're crawling in this space, so we're de-risking by flying over very sparse locations and then only coming in at the end over populated locations when we're actually doing the delivery. It was very successful for us. We were there for six, maybe eight months, and really, really positive, and it really showed us that there's a genuine need for this technology and a want for this technology.

Oranmore: a bigger town, new vendors, and coffee

[00:09:47] After Moneygall we progressed to Galway, into a town, Oranmore, a bigger town. Again, lots more to learn operationally, and also slightly different weather around the coast, windier, so really some interesting challenges there. Ten thousand people and about 3,000 homes. Again you can see we're based here in the center and we've got these highways that kind of go down.

[00:10:16] The big difference in Oranmore versus Moneygall, apart from the population density, was that we actually onboarded a load of new vendors. Tesco was one of our kind of flagship vendors that we started working with there, and we were actually operating the drones off the roof of a Tesco, the Tesco store in Oranmore, which again is a challenging environment, quite a small location where we were landing and taking off drones, and that required a good bit of work to ensure that we could do that. Again, everything comes into safety.

[00:10:50] Interestingly, when we were down in Oranmore, that's actually when we developed our first web presence. We had a mobile web, a mobile optimized page, for people to actually make deliveries and order, and they were full customers. What we discovered in Oranmore is that coffee is our number one product. Nobody thought that they could get coffee delivered, and we can deliver the coffee with the coffee art, as people call it, on top and with the foam still intact and everything like that, and we can get that to people in under three minutes. So again, a very, very compelling option for people.

[00:11:28] This is just a zoomed in view so you can kind of see the density of the area that we were working in, and again each one of these orange dots is a real delivery. We were in Oranmore for about a year and a half. When we were in Oranmore we ran a lot of feedback from customers, and one of the things we get asked a lot is, do people actually like this, or do they mind us flying over them and whatnot? On the survey we did, 98% of people are hugely positive. Particularly during COVID, I think people were looking to get things delivered and it was a great time to kind of launch into full-scale operations.

[00:12:07] It's amazing how normal it becomes in a town, and actually it was very hard to leave both Moneygall and Oranmore. But it's important for us as a business that we move on and we face the operational challenges of bigger locations. Again, feedback from customers hugely positive, and that continues today where we are in Balbriggan.

Balbriggan: the current operation

[00:12:32] So that brings us to our latest town. We've been there since January 2022 and we've done upwards of 5,000 deliveries and nearly flown 10,000 kilometers to make those deliveries, and again all of that is green, safe, with autonomous flight. Each one of these green lines is a delivery, and each one of these dots is the density of those deliveries.

[00:12:54] What you'll notice here versus the other kinds of, Oranmore and Moneygall, is that we're flying much more as the crow flies, and this is because we've now put 120,000 delivery flights on the clock on the technology, and we're building confidence along with the Irish regulator, who's building confidence. We still don't go exactly as the crow flies, because we are very conscious of the ground risk. So hot areas that we know will be highly populated, such as schools or other congregations, we actually avoid those, using a mapping system that we built to enable our aircraft to deconflict each other in the sky. What that means is that we can fly multiple aircraft at a time and ensure that they don't collide or get in each other's way.

[00:13:45] As I said, we're live in Balbriggan and we fly a weight of two kilograms, and we fly that into a two kilometer radius. So 35,000 people, nearly 10,000 homes. We're doing that week in, week out since January, every day Wednesday to Sunday, eight in the morning to nine at night, and it's really, really popular. We've got the likes of Subway, Tesco, and coffee obviously our number one, Coca-Cola, Advantage Aries[?], and then we've got different hot food vendors in the evening for pizza and Chinese takeaways, different things like that.

Watching a live delivery

[00:14:23] So what I'm going to do now is just show you a quick video. Here we have our live operation in Balbriggan. This is actually filmed in one of the early months in public when we first arrived, and you can see our aircraft there on the pad. So that day we were flying two aircraft, and again you can see you've got an operator down here and they've got control. For regulation at the moment we need to have visual line of sight of the aircraft, which is why we work within a two kilometer radius, and those pilots have controls that can interface with the drone should they need to. But in a typical operation they don't interface with the drone at all. A mission is loaded onto the drone automatically and the drone takes off, as it's doing there. Prior to takeoff you would have seen another operator comes and actually loads the drone, which is actually just putting the coffee or payload into the aircraft.

[00:15:19] So the aircraft has taken off, it's flying at 50 meters, about 80, 85 kilometers an hour, and you see it'll come in now into shot, I think it's up here. And then what you see is, this is again in the early days of Balbriggan, there's quite a bit of fanfare, and it's amazing, after about three weeks all of that dies down and this just becomes normal and people don't even look up. You can see the aircraft coming in here. So what it's doing is it's coming in above the drop zone and it'll come down to about 15 meters, and then you'll see the doors under the aircraft open and the payload will come down slowly on a thread.

[00:16:29] And there it goes. That was actually a cup of coffee and I think a Danish pastry. And that's it, no involvement from any users. The craft doesn't land, for obvious reasons. It makes its delivery, doors closed, it goes back, returns to base and is ready to go again. So again, a very simple process on the face of it, and a very compelling product for the customers.

Fun facts, and the delivery flow

[00:17:07] So that was a live delivery, and again we're doing that every day in Balbriggan, Wednesday to Sunday. There are some fun facts about Balbriggan and about the operation so far. We've delivered 500 liters of coffee. We've delivered to one in six homes, but actually it's now one in four homes. 115,000 flights — again, that's gone up since, 120,000 flights around now. Our fastest ever delivery was 60 seconds, and that was somebody who lived very close. Eight times more efficient, and really, really strong customer satisfaction scores.

[00:17:41] We are already seeing this in Balbriggan and the other towns, as to how this is changing behavior. Who would have ever ordered a coffee before this technology, because the quality wouldn't be there? We're seeing people order — I think the oddest order we've had is nappy cream, and a head of broccoli at eight o'clock at night. So people are really using this for emergency items and last minute items, and it's great to see the different use cases that come in for the technology.

[00:18:18] Just in terms of the delivery flows, just explaining exactly how it happens and typical timings. An order gets created and it moves through different statuses, so accepted, and then the store accepts it. We've had to build technology at each of these points, so our vendors have apps and devices to accept orders and manage the order, and our controllers have devices again to manage the order and figure out where it is. We've also then had to build a customer experience, so in Balbriggan we launched with a dedicated mobile app, so we're now app only, and that's great because we can give the customer feedback on the order, exactly as to where it is, very much like you get from your traditional aggregator app: your order is now being prepared in the kitchen, and it's being received, and it's ready. And the difference in our app, instead of seeing a driver on the road, you actually see the aircraft flying, and it doesn't get stuck in traffic, which is very, very useful.

[00:19:20] This is also very useful from an analytical point of view, because we're now able to feed this data back into our vendors to let them know where potential slowdowns or optimizations or capacity planning can happen in their own chain, which is very, very insightful for them.

Regulation and what comes next

[00:19:36] So to round up. Very exciting journey we're on and we think we're only at the beginning of it. We've got our location now in Dublin with 35,000 people, and we're looking to expand that to one or two sites across Europe, and also into the US next year. One of the contexts that we have to work in is regulation, and that's, believe it or not, a really interesting angle on what we're doing. Obviously we're flying objects in the sky, a heavily regulated space, and up to about two years ago the only regulation in place was that of manned aviation, that you would get on your typical commercial airliner, which is a huge, huge burden to go through.

[00:20:26] Europe and the FAA in the US are working on slightly different pages at the moment, but they're both actively trying to get, and succeeding in getting, unmanned aircraft certification programs in place. So Europe is a little bit ahead. What those programs mean is that we will be able to expand our operation with a more cost friendly approach. We expect a lot of news to happen in 2023, and one of the things that will happen is that we will be able to reduce our reliance on operators, where it's a one-to-one ratio, and we can move to a higher ratio of drones to operator. And again, all of that leads into autonomy and bringing the cost down per delivery. So very exciting times ahead. We're going to be around UXDX for the duration, so please stop by, we'll have an aircraft on site that you can have a look at. So thank you very much.

Q&A: hardware as a means to an end

[00:21:27] Host: Thank you Alan. I really enjoyed that, I thought that was really interesting, the different challenges that people have to face when iterating with the hardware side of things as well as the software.

[00:21:38] Alan: Yeah, it's definitely a challenge. It feels funny to say it now, but we envision ourselves as a software business, and software is the major problem here. The craft is brilliant and it's so exciting to build and it's really, really interesting, but it's a means to an end. A lot of the thinking is going into the software, the logistics platform, and how do we tell the aircraft where to go. It's reasonably straightforward — and I have to say that very carefully, because it's on my team, don't lynch me — but it's reasonably straightforward to get something to fly. What's very difficult is to get something to fly reliably, and that's all we've been doing over the last three years. And it's even more difficult to get multiple of those things flying reliably together, and that's where the big software piece comes in.

[00:22:38] Host: Yeah, so I might actually jump straight to that point, because you mentioned it there, the reliability and the multiple in the sky. I can see it as a challenge and I'm not going to downplay it, that if you've got multiple of your own drones in the sky it's hard to make sure that they don't crash into each other, but you could probably have some communication. But what happens when — because Google have Wing and Amazon are doing their deliveries, and you've got those guys in the sky, and then maybe some hobby drones or helicopters or other items in the sky — so how do you make sure that you're keeping it safe, not just from your own drones but from everything else?

Q&A: airspace coordination and UTM

[00:23:17] Alan: Yeah, great question, and it's one of the biggest challenges, it's actually probably the biggest challenge in the drone space at the moment. If you assume that the hardware is a given and we can make that reliable, it's that airspace coordination that's probably the number one reason why we're not seeing a lot of drones in the sky, or the number of drones that we should be seeing in the sky at the moment. They call that problem UTM, which is unmanned system traffic management, and it's sort of the equivalent of air traffic control problems.

[00:23:47] Alan: Everybody's familiar with air traffic control, so above 500 feet everything is very much controlled by air traffic controllers in each jurisdiction. And we think there's a lot of aircraft in the sky in terms of commercial aviation; in reality, what's going to be coming with UAVs, or drones, unmanned aircraft, there's not. So everybody realizes that the manual effort that air traffic controllers have to do now isn't feasible for unmanned aircraft, and so they're trying to solve this problem, which is UTM. There's a lot of companies at the moment now spinning up to solve that problem.

[00:24:27] Alan: But really, regulators are pushing for that to be solved in a federated way. So in terms of, we end up working with a lot of what people might call our competitors, but we actually all work together to try and solve that problem, because we all want the skies to be open. That's the overriding goal here, that we want the skies to be open for business, and working together will enable that. So in some ways we don't compete with each other, we're all competing for the same prize, to get the skies open.

[00:24:54] Alan: The way we do that is through working groups, and there's a great standard out, published last year, an ASTM standard, which actually describes how that interoperability can work, so we can tell the likes of Google Wing where our aircraft are and they can tell us where their aircraft is, and you can do that in a federated way. It would be quite easy to just implement one system where somebody owns the airspace in a particular country, but nobody wants that. Nobody wants a monopoly, everybody wants to do it in a federated, fair way. And it's, how do you make the sky fair? So if we're doing a delivery and Google Wing are doing a delivery, who has right of way, who has preference? And then if an air ambulance comes through, obviously that has preference.

[00:25:44] Alan: So all of these things are being figured out and worked out, because as of last year there's a specification in place that people are working towards, including ourselves, that will allow us to actually do combined operations with other drone companies. They could be survey companies, they could be delivery companies. So many use cases for this technology, but it all boils down to that same problem of how do you coordinate and interact with each other. It's a really, really interesting space, it's kind of coming to life now, and I think there'll be some big announcements. There's been a lot of trials around Europe, and there's a new initiative called U-space, which is a kind of designated set of rules that you can put into an airspace, and it's all with the aim to get to an ultimate UTM solution. So there's been a lot of those trials around Europe, there's I think trials happening in Ireland over the next two years on that technology as well. So all of that stuff is coming.

Q&A: dumb things in the sky, and detect and avoid

[00:26:40] Host: Great, that sounds like really promising. But what I was thinking when I was listening to that is, that relies on, I guess, smart things in the sky, and I'm thinking there's probably a lot of dumb things in the sky. What I mean by that is, if I'm just manually operating my drone, because that's manually operating I won't have this kind of communication that something else is there. Or, like, just animals, there's birds in the sky, they're not going to be telling your drones where they are and things like that. So how do you make sure that you're not going to crash with, I guess, kind of animals like birds, but also the hobby drone?

[00:27:22] Alan: There's two ways to look at it. They're just going to start enforcing a rule that they've had for about two years in the US, the FAA, called Remote ID. Remote ID is a set of technologies that broadcasts the location of a UAV, and it does that over long-range Bluetooth[?] or Wi-Fi or even the LTE network. That was supposed to be enforced in September of this year, it's now December 16th, so every drone now in the US has to have one of these Remote ID devices. Essentially they call it a license plate for the drone, and what you can do is you can get an app on your phone and you can hold your phone to the sky pointed at the drone, and it'll tell you on your phone the location of the pilot or the system controlling it, and the location of the drone, and who owns it and who to contact if there's a problem.

[00:28:13] Alan: So that's the first kind of big step around understanding what's in the air, and that's great and that's super cool. If you were guaranteed that 100% of everything in the sky would have a Remote ID, you wouldn't need any other complex solutions. But the reality is, how do you enforce that? How do you ensure that a drone doesn't take off without this Remote ID, or a bird or some other creature just in the sky?

[00:28:45] Alan: So that's where you get into what they call strategic deconfliction, which is basically you take a snapshot of the sky at a point in time just before takeoff and you plan your route based on strategic things like, I know this is going to be here at this time, so we can go here, et cetera. And then there's the tactical deconfliction problem, which is, I'm in the sky and something's going to happen that we don't know about, and that could be somebody has got a drone for Christmas and they throw it into the sky, not realizing that there's consequences. And then you're down to what sensors do you have on board.

[00:29:22] Alan: So we would have multiple sensors on board our aircraft, from vision, lidar, radar, to detect these things. And unfortunately there's not one silver bullet technology that can actually do that yet, and there's again a lot of companies working on this detect and avoid problem, and there's pros and cons with all of the technology. Lidar doesn't work great in rain, vision doesn't work great at night, radar tends to be pretty heavy, which isn't great for an aircraft. So there's all of these trade-offs being made. But again, there's lots of things happening in that space, and we've done a lot of work with AI this year around computer vision, and particularly around figuring out landing zones, but that technology can be augmented to detect things forward-facing in the sky. So there's lots of interesting things happening.

[00:30:15] Alan: And certainly in Europe now there's regulations in place. There's kind of a rule book laid out in Europe now, and if you follow that rule book you can fly beyond visual line of sight and with multiple operators to multiple drones. So what we were missing about three years ago as an industry was that rule book. We now have that rule book. The previous rule book was manned aviation; we now have a specific UAV rule book, which will allow us to set realistic timelines for when things will happen, and we believe 2023 is going to be a huge year.

Q&A: unit economics

[00:30:53] Host: Yeah, so I wanted to actually jump in on that, because at the very start of your talk you said that the economics doesn't work for kind of most deliveries, and I think you threw out the figure of ten dollars as the average kind of cost of a delivery, and a lot of that was down to labor. But you then said that you need somebody there looking at the line of sight for everything. So where are you at with your economics, and obviously it's a new industry so I'm assuming you're on a downward trend, but where do you think you can get to with that?

[00:31:30] Alan: Down to a dollar, from the current ten dollars. People pay, let's say, five dollars for a delivery now, and the other five dollars of that ten dollars is being eaten by the restaurant or the vendor or whoever's sending the goods. So last mile delivery there is loss making, and you hope that you can recoup that based on the product you're delivering. And our unit economics at the moment are certainly not positive, and they won't be positive until we can fully automate our solution.

[00:32:06] Alan: Now, we have all of the technology done to automate our system, and so in our test site, for example, we fly multiple drones to a single pilot. But that technology can't be rolled out in a live environment, because there's no regulation to allow that yet within Europe. We could manually take what we're doing in Europe — we've been awarded a license, an LUC, a light unmanned aircraft certificate[?] — so we can technically do what we're doing anywhere in Europe, but it wouldn't be beneficial to Manna because of the current unit economics.

[00:32:45] Alan: So what we need to do is we need to go through the regulation framework that's been laid out, which we're in the middle of, and once we have that done we can start then optimizing our operation for full autonomy. So we've got a very clear path as to how we drive our unit economics to where we want them to be, and once that happens, that's when we can scale. That's the reason that we don't stay, we haven't to date stayed in a town, because it's all about learning and then pushing us on to the next milestone and the next milestone. Now, Balbriggan is a great case: 35,000 people, it's the second most densely populated town in Ireland, so it's a great location for us now and probably long term.

Q&A: noise

[00:33:28] Host: Awesome. And just one on the density actually, because I've heard people kind of flying the hobby drones around, they can be quite loud. Is that an issue that you have, with noise of your drones in a densely populated area?

[00:33:43] Alan: Yeah, noise is probably the biggest question we get. What happens is, the first time people see our aircraft they say, it's bigger than I thought it'd be. And then we explain it has to be this size because we're fitting pizzas. And the second thing they say is, it's quieter than I thought. So everybody has a really, really negative connotation with drones, because the little small drones are incredibly loud and annoying and it's just not a pleasant sound.

[00:34:13] Alan: In the picture behind me, the propellers on our drone are quite substantial, they're 28 inches, and what that means is actually they're very quiet. Typically the bigger the prop, the quieter the drone. And we're also flying at 60 meters, so you can't hear us when we're at 60 meters. You hear us when we come down to deliver and there's noise, but that noise is no more than traffic noise which you would get from a normal road-based delivery. So it's a very hard one to demonstrate, because video and microphones on phones and cameras don't do it justice. But it's the second thing that people say about the drone, and we have people visiting us all of the time from different drone businesses and around the industry, and it's the one comment that we consistently get: how did you make the drone so quiet?

Q&A: payload and future aircraft

[00:35:10] Host: Awesome. One last question. Two kilos, so that's your current limit. Is that it, or do you see in the future you might have a 10 kilo drone, or is there a point where it just doesn't make sense, like you wouldn't be delivering furniture by drone?

[00:35:27] Alan: Yeah, no, I think we're a long way off delivering furniture or fridges or ovens or anything like that. The reason we're two kilograms is, we have to keep the drone, for regulation reasons, below 25 kilos fully loaded up, so that includes your battery, the airframe, the craft and everything like that. Our craft could technically carry a lot more than two kilograms, but we wouldn't get the two kilometer distance in all of the weather conditions that we want to fly in. So the weight you carry is a trade-off between your performance envelope and what you want to do.

[00:36:03] Alan: Most people would only fly their drones in good weather, because they're sensible, and they're standing there flying their drone. We're looking at it that our drones are autonomous so they don't really mind the weather, so that introduces a load of design challenges to the aircraft to make it perform. Our future aircraft is going to have a much bigger payload capacity, of three and a half kilos, and a bigger volume. Volume is actually our challenge, so at the moment we can fit in about two, two and a half shoe boxes kind of volume, and the next aircraft is going to be significantly larger than that, and so that's going to allow us to do more and fly more in different conditions.

[00:36:49] Alan: At the moment, if somebody orders, let's say, a really large order that's four kilograms, we'll actually do that in two flights. It's still way more efficient than road-based delivery, and we do that because we want to prove market fit for an extended size. But I think definitely in the future, the aircraft, as I said before, is an enabler, and not all aircraft suit every situation. So if you're delivering a cup of coffee you might want a slightly smaller aircraft; if you're delivering 20 cups of coffee you might want a bigger aircraft. The future is definitely multi-aircraft.

[00:37:33] Alan: But as a startup we need to get one craft working, which is kind of middle of the road, and then in the future — a lot of the technology that we're building, nearly all the technology that we're building, is transferable across craft. So all of the work that we've done, for example, on the triplex avionics system, like all of those electronics and software, we just lift that off and we can put it into any airframe we like. And that's hugely important for growth and scale, and as we enter different markets with different needs.