The Economics of Drone Delivery | Hacker News Reader

The Economics of Drone Delivery

flexport.com

76 pointsFuturebot45 comments

Strange that an article with this title that talks so much about Amazon's drone efforts doesn't examine Amazon's relationship with the delivery industry. Amazon desperately wants to economically remove the time lag between buying something from Amazon and buying from the nearest Wal-Mart. Drones have the potential of significantly improving delivery times and costs in the long run, but the delivery industry has little incentive to innovate in this area since they already command high margins on small package deliveries.

For Amazon, it's not necessarily about how well they can execute on their drone delivery program. I imagine that Amazon will actually make a go of turning drone fulfillment into a business, a la AWS, though the effort is still worthwhile for them even if that never happens. With this drone program, they are able to provide a large scale proof of concept, working through legal barriers that are relatively trivial for a company with Amazon's resources, but insurmountable for hypothetical drone logistics startups that might want to put pressure on UPS et al. But once the grunt work is done, the barriers to competing in the logistics industry through drone fleets will be substantially diminished.

This is the big upside for Amazon here: paving the way for increased competition in the delivery industry (whether it ends up primarily being driven by Amazon, incumbent companies, or startups) which will have the likely impact of improving it's unit costs while having the potential of eventually bringing it into parity with regards to the largest remaining advantage that Wal-Mart retains, same-day delivery.

The unit costs and limitations aren't particularly relevant to much at this point in time. Those will significantly improve as competition increases, as is typical in maturing technology industries.

Whether or not drone delivery becomes a mainstream part of the equation--I suspect it won't but that's beside the point--it's certainly in Amazon's interests to research and encourage innovation in the logistics business. Mind you, there's already a lot of innovation happening in supply chain systems but economical 1 hr to 1 day consumer delivery hasn't had any real breakthroughs.

Now that may be because people aren't willing to shoulder the cost for most mainstream use cases. But in that case, it's worth experimenting to see if costs can be driven down or use cases can be expanded in various ways.

Cynics might suggest that even if situations where drones are the most commercially viable delivery solution prove to be limited, their putative threat adds to Amazon's negotiation power with their contractors...

I've got to think that Amazon could rent out the roofs on commercial buildings throughout cities pretty cheaply. Doing so could completely moot the range issue as it would enable drone relays. I'm sure there will be regulatory barriers to doing that, but if they can get FAA approval for their delivery program as a whole, I'm sure they can get to a tipping point of municipalities that enable this sort of relay system. It might even be a Google Fiber sort of thing, where people actively lobby their municipalities to remove regulatory barriers so as to bring the city's technological capabilities in line with the state of the art.

Agreed. Another possibility would be a hybrid approach with delivery trucks acting as mobile distribution centers. Trucks could be loaded with drones and packages, park in rural areas and let the drones take care of the last mile.

That's precisely what Matternet (mentioned in the article) is planning! An internet...but for Matter. Sorry.

I wonder how energy costs would factor into these economics. A couple of thoughts off the top of my head:

* drones run on electricity - they are less energy efficient than ground vehicles, but this can be offset by cheapness of electricity - particularly in areas where battery recharge could be done by solar or wind.

* The large banks of batteries required to make the drones economically efficient (I imagine it would be cheapest to have a couple sets of batteries per drone and swap them out to allow continuous drone operation, rather than recharging each individual drone) might be useful for secondary operations when delivery periods are slow - e.g. using the bank of solar charged batteries to power night operations when the next day's deliveries are going to be slow.

Similarly I wonder how having an army of drones out in the world might have secondary economic impacts for the operators:

* For a small power budget, the drones could be equipped with additional sensors to gather data on various things and provide a sensor stream for 3rd party purchase.

* the video information I would imagine they already stream back could provide all sorts of benefit - selling the video to the news if a drone is in the area of a newsworthy event.

* similar to the previous point, but for traffic info.

An airplane is far more efficient than a helicopter (drone). Also, I can't imagine that a drone flying the last mile is going to be less energy efficient than driving that big heavy delivery truck.

Per pound of cargo I'm sure the delivery truck is way more energy efficient if it's relatively full.

Per pound per mile, I'm sure it is. But most packages get driven way farther than they need to be, since one truck is out all day dropping off packages. Per pound per delivery, I'd bet the truck loses.

That's a good point, and a lot harder to estimate since there are probably a wide variety of routes that a package might take.

Interesting. The counter-point to this would be that the drone needs to take a return trip for each package, whereas the delivery truck is able to "batch." Not sure which effect dominates.

An airplane has a better chance of gliding out of harms way as well whereas a helicopter jut drops.

Thinking a little further. In the future, we will have autonomous cars, and a user will probably not "own" the car, but will simply order it. And the user can specify if (s)he would like to share their ride. This car can easily have additional functions, for example package-delivery (a package is nothing more than a shared ride with a non-human passenger). Of course the logistics will be run from a central computer system. Perhaps this could be a more cost-effective solution to the whole package-delivery problem (?)

Yes, and package delivery autonomous car can be made very small -- just big enough to carry the package. In fact, a ground-based autonomous delivery vehicle doesn't have to look anything like a car. Imaging an autonomous Segway-like 2-wheel cart. Segway is 4.5 times more efficient that Toyota Prius and 11 times more efficient than average American passenger vehicle.[1] Something like that would allow delivering packages very cheaply.

[1] http://www.segway.com/downloads/pdfs/energy_efficient_segway...

If everything is really well orchestrated logistically, your Segway solution could attach (piggyback) to a bigger autonomous vehicle for a certain duration for greater efficiency, higher speed and bigger radius. Of course, the fuel costs will have to be split, but that is just a matter of making smart accounting software.

I like that idea -- for less time-sensetive deliveries you could have a "mothership" truck loaded with mini-vehicles drive to a neighborhood center and then release them. However, for things like food delivery where you want to deliver in under an hour, it will be better to just use mini-vehicles.

What we have here is kind of lack of imagination. There are at least 2 parts of delivery that I see can be worked on with drones. One is last mile delivery, that honestly is challenging, unless you live in suburban or rural area.

However one, more interesting and probably lucrative part that should be easier to implement would be transport between major hub to nearby delivery point. Since these points can be built out and/or prepared, you can have deliveries coming non-stop, through FAA approved route, and then delivered from there. It doesn't even have to be dedicated Amazon employee, they can offer pick up or delivery by Uber or something like that.

For me this part can be very creative and I am surprised that Amazon doesn't see this as promising area of investment.

That is certainly an interesting point. Modern distribution centers use conveyers to move packages to the truck loading port where they will be loaded on to trucks for delivery. Why not extend the conveyer through the sky to a truck loading port a couple of miles away. If you lay out reliable ingress/egress channels for the drone traffic you could probably stack 300 - 400 drones per mile in that channel, operating all day would certainly expand the virtual "size" of your warehouse.

That is pretty much how I would envision it. Once you build that, suddenly another level of things that we can't even imagine today are becoming available.

Plus ecologically would probably be very good and less trucks on the roads.

The problems with the major hub to the delivery point are the same as some of those identified in the article: drones currently in operation don't have much carrying capability, and compensating for that with significantly more drones adds cost over a truck. Even if you scale your drones up to truck size and don't have any range limitations, you're carrying packages a similar distance whilst fighting gravity all the way. And if bottlenecks with sorting at each end and last mile are unchanged, the extra cost of cutting out trucks sat in traffic in the middle of the supply chain might make surprisingly little distance to delivery times.

Getting the FAA to approve airways for long distance autonomous commercial cargo transport without prohibitive fees sounds like a challenge too...

So what you are saying, there are no serious technical challenges (I think there are few challenges to be solved), just bureaucratic ones. That one are always solved with money, something Amazon clearly has.

Again, this is just an idea to show how this can be much more than it is today.

That's not really what I was saying at all. I'm saying we have good reason to expect drone technology to continue to be vastly more expensive than trucks for transporting groups of packages over long distance, and little reason to believe that drones will make a significant enough impact on delivery time for it to be worth it for the types of products normally ordered from Amazon since the bottlenecks are all elsewhere.

I haven't seen too many long-distance, high-payload drones on the market either...

I suppose it depends on ones definition of "high payload". How many books to a Hellfire missile? Granted the military drones used for attack missions don't do VTOL, and aren't ready for autonomous take-off and landing (yet). But I believe there's lot of research in the area, and it doesn't seem inconceivable that a company like Amazon might get use from short, purpose-built landing-strips for similarily purpose-built drones.

Speaking of military delivery - I wonder if it might be possible to build one-off "drones-as-packaging" with maybe a parachute for breaking and a quadcopter-like design for final steering that allows for dropping "smart" packages from the sky down to a porch...

Suggesting a $5M piece of military hardware capable of carrying a single 100lb missile pretty much emphasises my point about trucks (or even motorbikes!) being far superior options for medium-long distance transit. No matter how much research goes into developing commercial alternatives, carrying books through the air over long distances at speed rather than driving them along roads will expend significantly more energy and cost more money, because physics.

And whilst getting goods by parachute might have novelty value if I'm around to watch or be extremely useful if I'm stuck somewhere up a mountain, for everyday goods delivery I'd frankly rather have them handed to me by a person that considers the goods "delivered" when they're in my hands rather than probably in the right garden. It's also why their current delivery drivers aren't encouraged to save time by throwing non-fragile items at customers' houses as they drive past.

Fair points. I meant that there's lot of research and proven designs in fixed-wing drones that are much more fuel efficient than quadcopters.

I'ld add that drone delivery for takeout would be a nice win, given that most restaurants that deliver are already limited to nearby locations and speed is desired.

With a proviso that it might be easier in a suburban situation to begin (given a lawn to drop off the goods).

The thing is that quick, low-price takeout delivery to a limited radius of households is pretty much a solved problem. I suppose one could imagine more industrialized takeout-only restaurants in suburban locations but, again,one could create such a chain today and indeed we have them for say pizza. They're not very good but they're big businesses.

If you could solve the problem of needing to directly interact with the customer to collect payment, I'd think food delivery would love the idea of drones. The biggest cost of a pizza delivery shop is paying the drivers. They normally get an hourly wage, plus mileage if they are using their own cars.

They'd need to solve payment collection; a lot of people still pay cash for things like food delivery.

I don't use any of them because I don't live in an urban area, but my understanding is that apps like GrubHub can handle the payment process.

This "They'd need to solve payment collection;" is an intriguing problem if it is still present. The spate of startups (doordash/postmates/et.al) solves the problem (and more) but I certainly wonder if there's a more focused solution to the specific problem. Would there happen to be more specific details/datapoints?

> Instead, the real challenge is the regulatory environment. The FAA has banned all commercial uses of drones in the U.S.

I always don't understand this reasoning. There are tons of other markets, many of which Amazon is active in. Being successful there without any problems would be the perfect argument to also allow it in the US.

Amazon has set up drone delivery research centers in the UK and Israel for just this reason.

what about delivery truck as drone carrier?

when the truck is close to destination, drone(s) are detached to drop/pick package(s), only to return to the truck after finishing the work. this eliminates truck stops.

imagine a flying bee hive

I would think that once you're sending out trucks anyway, there isn't much even potential win in layering a complex air delivery system on top of that. In addition, you're adding another source of latency and, to the degree that drone delivery makes any sense at all, I expect it would be used for very time sensitive material.

But drones are much more range- and speed-limited than trucks are. It might very well make sense to drive a truck from New Jersey to Manhattan, or from Irving to Plano, and then let the drones spread out.

We know that much of a delivery network's cost is the last mile, but I'd be interested in the costs of the last 100 yards. Suddenly, instead of one driver/one truck per neighborhood per day it's one stop per neighborhood per day. They can stop once or twice in the neighborhood, load drones with packages and then drones do all the deliveries. If you even have a separate "loader" in the back, you can have the driver jump out and make the few targeted heavy package (20+lbs) deliveries.

Maybe not trucks specifically, but I would imagine that having the notion of mobile or temporary drone launch points would help a lot during busy times - e.g. the week before christmas.

I watched all the major carriers sending out multiple delivery dispatches a day through my neighborhood. To the point that the bottleneck appeared to be concurrency issues on driver time (as I tracked a few packages through the various systems). For my family, many of the packages fell within reasonable drone parameters and a lot of them had a significant pause time at local delivery base (e.g. UPS truck loading center or post ofice). I imagine a steady stream of drones from a temporary drone base would help lighten this load. Imagine sending a semi-truck that operates as a drone base to a big parking lot, appropriately located to be in a delivery dense area. A handful of employees does piloting and battery swapping and drone loading, while a delivery truck brings packages to the temporary hub. When the busy and/or overload time is over, the truck can go back home, or on to the next place expecting an overload. It wouldn't take a lot of saved overtime and seasonal labor costs to make this a win.

edit: I would also imagine that a lot of driver time during christmas anyway is wasted on traffic issues, part of which are compounded by them! Drones wouldn't have this so much.

Organ delivery might be a good place to test initial drone delivery. It's very high value with very few, well-defined pick-up and delivery points.

Driverless trucks as distribution centers/battery swaps, and rooftop drone stations, could bring this down to last meter delivery, rather than last mile delivery. You just need delivery from curb to door step which a truck can now do and multiple drones can be delivering packages at one time.

Going to back-of-the-napkin this here: Current cost (just for person making the drop) is about $1.50/stop. Assume six packages/stop. So now you just need six drones to fly a single package from a truck to the doorstep for $0.25/each in that stop. So each drone will need to make 120 such runs/day for under $30.00 ($150/week, $7,800/year). This should be enough to easily cover energy/CAPEX as I assume that drones would be operating autonomously, and maybe you'd need one override-pilot in a support center for every 100 drones (basically one manual override every 17 stops).

Given the sanity check above, I'd wager that you could easily get this down to $0.05-$0.10 per package, including edge cases covering 99% of the deliveries. Big deliveries like TVs may ask the owner to come out to the (driverless) truck to pick it up at the side of the road while a drone century waits.

The killer apps for drones are smuggling and delivery of illegal drugs. A kilogram of drugs is very expensive, so the drone costs are not relevant and it allows the seller not to be caught. They will also not be stopped by government regulations.

But it also runs afoul of the "don't break two laws at once" principle. If you deliver drugs via conventional methods, the police need probable cause to stop and search you for drugs. If you deliver them via illegal drone, they can stop and search your drones whenever they want.

Drones are autonomous. Trucks require drivers. So the competition is between how quickly trucks get automated vs the total system capacity of drones for delivering stuff.

UPS cost != UPS price

Also, not taking into account cost of roads that UPS doesn't pay, nor the fact that flights are much more flexible compared to building roads

UPS pays for roads. They pay income tax generally, and they pay road use tax on all their vehicles, and they pay for fuel which also has road tax.

Reply on news.ycombinator.com