Yesterday we got a little glimpse in to the sort of things that make Amazon CEO Jeff Bezos' pants tighten, and it's all flying robots dropping boxes. Bezos' drone-delivery system, called Amazon PrimeAir, was introduced with a little video showing how he imagines the service to work. I'm here to tell how it could really work.

The video is pretty straightforward, showing a drone making a sample delivery, but the details shown in the video don't really seem to square with how such a service could actually work (consistent with the theory that this is just brilliant Cyber Monday viral marketing).


The range of those drones can't be huge, and yet the video shows the drone leaving an Amazon warehouse, crossing what seem to be vast acres of empty, rolling green grassland/golf course, and then happily plopping the package down in front of the door of the expecting customer.


That customer is also waiting inside the door, looking outside with an odd combination of fear and eager expectation, like he would if Bezos' plan was to deliver by barely-trained direwolf.

Here's the basic issues with what's shown in the video: the area shown for drone delivery hardly seems like the ideal for this sort of service, unless the customer lives off a golf course that abuts an Amazon warehouse, there seems to be no regulation or control of where these drones could go or fly, just dropping the package down in front of the door wouldn't work for huge numbers of people, and lastly, that plastic lunchbox thing.

So let's go through what it would take to make this actually work:

1. Location. The suburban-ish looking area shown in the video doesn't really make sense for this — the distances are too vast, and conventional trucks would likely be more efficient. Where this makes sense is in big, dense cities.

Big, dense cities have enough customers in a small enough area to make this sort of thing economically and technically viable. The range on an octocopter-like drone shown in the video isn't huge. A similar model from Steadidrone lists a 15 minute flight time, depending on load. Even giving Amazon the benefit of the doubt and doubling this time, even with a full load, that doesn't give a whole lot of range — probably 5 miles or so at best. Big cities like New York, San Francisco, London, Tokyo, etc. are the way to go.

This, of course, means multiple smaller warehousing hubs inside of these cities, too, but that seems pretty achievable.

2. Drop-off points. Big, dense cities also usually mean lots of Amazon customers who live in apartments. Which means unless a drone can also open a door or talk to a doorman and manipulate elevator buttons, it's not dropping squat off on your doormat.

Another system is needed, and it's something we all remember from kindergarten. Cubbies.

The rooftops of buildings that Amazon drones can deliver to would have rows of cubbyholes for each of the building's apartments. Inside each cubby would be markings to allow the drone to properly orient the package so a QR code or similar marking on the lower part of the package would align with a small camera/sensor to read the code.

A small embedded computer — think Raspberry Pi type of thing, in a weatherproof box — that manages all the cubbies would note which cubby got a package, confirm that the address on the package matched the cubby's owner, and then send an email or text to the registered owner of the cubby. The owner could then either get the package themselves, or have the building's staff get it, depending on the building's own rules.

This would allow for simple delivery with notification to all apartments in the building, and be more secure and safe than just dropping it near a building's door.

3. That plastic lunchbox thing: Get rid of it. The cardboard boxes Amazon normally uses should be plenty strong and durable to handle a 30 min drone flight. Besides, what is the customer supposed to do with that plastic box as shown in the video? Send it back to Amazon via another drone? The original drone already left.

Maybe collect them? They look handy for keeping screws and stuff in, but after a few months worth of ordering, this is going to be a pain.

4. Flight paths. The video just showed the drone happily meandering to the customer's house, probably while whistling or stopping off for a corn dog or something. We all know that's not going to fly. Metaphorical "fly."

Cities have wires cell towers and people and birds and all sorts of stuff in the air, and above that helicopters, and above those airplanes, and all these things are much happier not being run into by a drone swirling eight tiny blades. The solution to this is that specific, flying-drone friendly areas of the sky need to be established.

In a big city, drone-friendly skylanes could be determined, and drones would be programmed to plot their routes using only these three-dimensional invisible sky-roads. They would avoid flying over areas of high congestion, avoid power lines and helicopter flight paths and all that.

It'll take some work to determine these, but they don't need to be very large at all, so I'm confident suitable lanes could be found.

5. Nets. Look, why not? Occasionally, something will go wrong. A demented pigeon will decide to take out a drone (I have heard a lot about angry birds in the media, and I'm assuming this is what they mean). There may be technical issues, or sudden bad weather, or any number of things that could cause either a drone to crash or at least drop its cargo. So, over sky lanes that cross heavy traffic or pedestrian areas, why not run some building-to-building safety nets?

Nets aren't that expensive, and even from a purely psychological standpoint they would make people crossing under skylanes of heavy drone traffic at least feel a little safer. If a net stops one accident or injury, it's worth it.

Plus, I'm sure Spiderman would appreciate the extra bit of insurance.

It does sound crazy and futuristic, the idea of flying robots dropping off Skittles and sex dolls and USB hubs right to you, but I think, properly implemented, it could work quite well. I'm just surprized Jeff Bezos didn't run it by me first so I could have cleared all this up initially. Do you think he knows how to get ahold of me?

Man, I hope so.