UK, Michigan & Sweden push robocars, Toyota doesn't -- and Amazon delivery drones


The past few weeks have been rife with governments deciding to throw support behind robocars.

I wrote earlier about the plan for pods in Milton Keynes, NW of London. The UK has also endowed a a £10m prize fund to build vehicles and for a town to adapt to them. This will be managed in part by the Oxford team which has built a self-driving Wildcat and Nissan LEAF.

In Michigan, they have been working on a new robocar law that may be the next one, and the University of Michigan has a plan to put a fleet of cars out by 2021. Ann Arbor is the site of the ITS V2V testbed, which will probably slow this effort down, but Michigan is keen on not having the auto industry taken away from it.

Volvo, while now a Chinese company, has had many efforts, including their Sartre convoy experiments. Now they have declared that they will have 100 cars on the road in Gothenberg in 2017. They will also build parking systems.

In spite of all this, Toyota recently declared it is only building vehicles for research purposes, and has no desire to market such cars. Toyota had been a leader among the Japanese companies (until Nissan took over that role by building a research lab in silicon valley) but it's surprising to see them drop out. Of course I predict they will regret that.

Amazon drone delivery

The big news this weekend was the announcement that wants to do drone delivery, accompanied with a concept video. This got everybody buzzing. I was interviewed for stories by the Washington Post and Wall Street Journal (paywall) as well as the New York Times because of my prior writings on deliverbots.

Some of you may remember I post I did early last year on drone defibrillator delivery and the efforts of our students at Singularity University to build Matternet for drone delivery in the developing world.

Drone delivery is interesting, though its big value will be in lightweight, urgent items like medicines. Ground vehicles will still win for cost and efficiency for most items. However, the drones can be much faster, and have options like delivering to places ground vehicles can't reach -- like your roof or your backyard. Deliverbots must get safe and legal on busy streets, drones have to figure out how to not hit one another (or people on the ground) in crowded airspace. The LIDARS that make ground vehicles practical have enough range for ground travel but poor range as flying sensors. Radar is good in the air but can have interference problems.

Getting a drone to land at any given address is a hard problem. There are trees, overhead wires, wind gusts and strange geometries. I suspect drone delivery will work best if the drop location has already been scanned and mapped. However, if there is a decent clearing, I could see it working by having the recipient put down a special marker (like a QR code) on the ground. GPS is not accurate enough to fly with but camera could pull out special markers.

One great marker would be your cell phone. Either with its "flash" LED pointed up and pulsing, or its screen, if the screen is bright enough. Go outside, put your phone down, have it guide the drone partway in with radio and GPS, and then have the drone's camera follow the flashing light. If phones had better raw GPS access (they don't -- not yet) they could also provide differential GPS information to a drone to guide it in.

This works because with robot delivery, you never need to deliver to an address -- you deliver to a person. Wherever that person is, or at least never when the person isn't there, unless you want to. A robot delivery service will wait for a signal that you are home or one the way before delivering to your home, but might also deliver to you in whatever parking lot you are in, or your office. The robot won't release the cargo unless it gets the ACK from your phone as you "sign" for it.

Multi-copter drones today don't have a lot of capacity and range, but it's improving. Liquid fuels for larger drones might help boost that. Fixed wing drones have much more capacity, but they need runways (or a skilled launcher) to take off. Some fixed-wing drones can land vertically if they have motors powerful enough to lower them down tail first though they tend to need something suitable to land on in such cases.

Robot delivery should make existing retailers, even big box ones like WalMart, scared of online retailers like Amazon. While a drone won't replace WalMart on a trip where you plan to fill your shopping cart, it might well be very suitable for the things you buy from Walgreens.


How large would a QR code need to be for a downward-facing camera on a drone to be able to properly resolve it from the air at 50 feet, or 100 feet, or more? Looking at it from the other direction, if a QR code is printed by a delivery recipient at maximum size on a sheet of 8.5"x11" paper, how low would the drone need to be for its camera to resolve it? Also, how far off from 90 degrees could a drone's camera reliably resolve a QR code on the ground?

QR codes have a fairly narrow range of "ideal circumstances," and I think a drone flying overhead scanning the ground for them would have trouble intersecting that range.

There are myriad other problems with delivery drones. For example, delivery via drone to people in multi-family dwellings is problematic, to say the least. Drones would work better in high density/urban environments due to their short range, though.

I expect there are laws in many places against deliveries to locations that don't have street addresses. So delivery to a person standing with a cellphone in a the parking lot of a business, or out in a field, might be prohibited. These laws might only apply to general goods delivery services, though, I'm not sure. Regardless, they'll vary across municipalities across the U.S. and the world, so that's a lot of legal weeds for Amazon to slog through.

Will the drones be flying after dark? In the winter they may not be able to avoid it, especially at higher latitudes. That means having marker lights. People might not appreciate lots of little lights flitting across the sky, especially in landmark/tourist-type areas.

Presumably there will be some laws against private individuals using drones to spy on their neighbors, peek through upper story windows, etc. If delivery drones become commonplace, though, individuals could just disguise their personal drones as delivery drones for "camouflage." Unfortunately this also applies to bad actors seeking to use drones as delivery systems for explosive devices, toxins, etc.

Deliveries by drones could also allow a person to get past a security screening or checkpoint and then have contraband delivered to them on the other side. (Perfect fodder for stories about spies and/or terrorists.) Again they could have a personal drone disguised as a commercial one to be less suspicious. Of course airports are going to be no-fly zones for drones in general, but there are other situations where this might apply, like outdoor festivals.

Will 3-D printing technology advance fast enough to make drone delivery a non-starter by the time drones are able to be deployed? 3-D printers and drones will have similar product size/weight constraints to start. 3-D printers will probably get higher capacity faster than drones will. The top value of goods delivered via drone is likely to be kept low, too, due to the risk of loss. So while 3-D printers are going to have limitations in printing high-intrinsic value things like gems and precious metals, you're not going to want to have those things delivered by drones, anyway.

There is a small QR code that would be reasonably readable from above, but I don't mean literally the QR system, but something that can be seen from within 20m or so, and has enough bits so you can't be tricked. The LED plan is much more visible, but the paper code can be used by people without a phone.

Multi family is not a problem in the situation I describe -- delivery is to you, not your address. So it is only delivered when you (or somebody you designate as trustworty) is there.

Delivery drones (and all commercial drones) might well be limited to corridors, which are over streets. (Though that has other problems with safety and congestion.)

Drone delivery of contraband will of course not bother to follow the rules.

3-D printing is cool but it's for one-offs, not mass produced items, at least not yet. If I want a taco by drone a 3-d printed one may not suffice.

If drones are going to use QR codes to deliver to an individual, no matter where they are, and that individual is out in the field, in a parking lot, etc. then I hope that person has a portable printer with them to print out a sheet of paper with the QR code to lay out on the ground for the drone to make the delivery.

At this point probably a lot more people have smartphones than still have printers, anyway.

QR codes for non-industrial use continue to be a solution in search of a problem. Delivery drone landing targets are just not something that they are a good solution for.

Multi-family units are going to be a problem even if the drone is delivering to just a person holding a phone. The drone needs to land somewhere. There may be no outdoors parking (underground/in-building parking). Roof access may be restricted. There's traffic out in the street. There are pedestrians on the sidewalk. Regardless, there may be a lot of trees creating a canopy over the sidewalk near the building.

"3-D printing is cool but it’s for one-offs, not mass produced items, at least not yet."

Aerial drone delivery isn't here, yet, either. The question is what becomes more viable, faster. Then even if drone delivery is viable for a while, how long will it be before 3-D printing seriously eats into the market for small items that would be delivered by drones? Aerial drone delivery may have a rapid peak and then a rapid fall, like digital physical media in the face of streaming and downloads (RIP Tower Records, Virgin Records, Blockbuster, Hollywood Video, The Wherehouse, Sam Goody, etc.).

Food delivery may remain a worthwhile option for aerial drone delivery, but it doesn't match up too well with independent, mom-n-pop type restaurants (high initial capital outlay, maintenance, insurance against loss, etc.). Sure there could be businesses that specialize in providing drone delivery services to multiple local restaurants, but the cost of operation is likely to be quite high, and the return from small businesses, who are already operating on slim margins, is likely to be low.

Besides, flying through the air for up to ten miles will have a significant cooling effect on the cargo, which would then need significant bulky, heavy insulation. This may not be too bad for a burrito or a couple of tacos, and actually beneficial if your order is just a pint of ice cream. :) For an extra large pizza or dense/heavy collection of enough Chinese food containers for more than just two people, though, it will be a problem.

I think 3D printing is exciting, but the problem is that a I look through my shopping history I see only a modest number of items that might be practically 3D printed. The main reason most don't qualify is that they have some electronics in them. We'll be able to 3D print simple electronics before too long, but the reason there are electronics in all these things is because processors are cheap, because they are made in big expensive fabs in large quantities. The day we can make a microprocessor in a 3D printer will be further away, but a different story.

Of course, one answer is that the 3D printer comes with a supply of general SoC microcontrollers which it can pick and place into 3-D printed circuits. Not going to be as cheap as picking just the right part but could be doable.

Currently there are no ways to make things akin to leather or fabrics or light things with tensile strength, and those are in a lot of products. The day will come -- but the drones probably come sooner. We already have the drones, the question is how to control them and legalize them.

Actually, when it comes to mom-and-pop restaurants, I predict the opposite. With robot delivery, now you can have a delivery "restaurant" in a nicely equipped home kitchen. Still need to be up to health code standards, of course, but much easier to start small. You could be a successful restaurant making just one dish very, very well. You announce to the aggregators, "I am making my paella tonight." If you're good, customers have given in top rankings, and your price is very good because all you do is collect the orders that come from the aggregator -- many of them early in the day from people planning an evening meal -- and order just-in-time delivery of fresh ingredients needed over your planned minimum. Then you make one giant batch, and dish it out into trays with preprinted labels, and place them in the robots that come to your door.

No tables, no waitstaff, no wasted food, all done by a single chef, it not being a lot harder to make 200 dishes than to make 20. Great economies of scale and delivery that is a few cents/mile. Can mom-n-pop food get any more efficient than that? Can any fresh food get much more efficient?

"We already have the drones, the question is how to control them and legalize them."

Ignoring the other practicalities like not necessarily having a place for the aerial drones to land once they reach their destination, and air traffic considerations, and security considerations of someone camouflaging drones carrying bombs, contraband, or surveillance equipment as innocuous retail deliveries... ;)

"Can mom-n-pop food get any more efficient than that?"

It comes back to the question of who owns the drones, and how much they're charging to make the deliveries. The initial capital outlay for the drones will be high, and there's costs for maintenance, loss, replacements, insurance (the drones would be insured against loss, there'd be insurance to cover accidents where a drone malfunctions or it's damaged and falls to damage property, people, or animals, and there'd also be insurance to cover failed or late deliveries), etc. In fact considering the legal liabilities, which would also include accusations of privacy invasion by drones flying overhead (they HAVE to have cameras in them to land), the drone operator would probably have to employ a full-time legal staff. The cost to the mom-n-pop restaurant or chef would likely be quite significant per delivery.

This is also ignoring the bulky and heavy insulation needed to keep hot food hot while it's flying through the air.

Plus there's the practical issue of how many drones can land, be loaded, and take off again to make their deliveries in a short period of time? A single chef may be able to make 200 batches of a dish at once, but how long would it take to package and load those dishes into 200 drones for delivery? There's also the logistics of making sure each drone delivers the right order, unless it's standardized to only one or two meals per drone. In this case you lose efficiency unless all orders are exactly the same.

Ultimately, I think the promise of aerial drone delivery may wind up akin to the dreams of ubiquitous pneumatic tube delivery systems in the late 1800's and early 1900's. In a similar way, then, the technology already existed, and it "just" needed implementation. (With similar handwaving at the time by its proponents about the obstacles. :)

"Robot delivery should make existing retailers, even big box ones like WalMart, scared of online retailers like Amazon. While a drone won’t replace WalMart on a trip where you plan to fill your shopping cart, it might well be very suitable for the things you buy from Walgreens."

Why would Walmart not be scared? Delivery drones in truck form can do a full walmart/supermarket type delivery from a warehouse that is much cheaper to run than a big box store and they don't have to fly.

Add new comment