Initially this will probably carry a significant premium over normal delivery. However, as someone who lives in a city, I can definitely see there being times when I would be willing to pay the premium. Taking a cab ride to and from somewhere can easily cost $15-$30, so it could actually be cheaper to have the item delivered by a drone if time is of the essence.
I'm saying that there is no reason for to be as noteworthy as it is right now. As Bezos himself says,
“this is early, this is still years away.”[1]
Yes, 30 minute drone delivery would be big news. That Amazon is in the very early stages of an R&D project to look at the feasibility of it is not. In any case, I am staking an early bet that you'll have your Amazon products delivered by a self driving van before you'll have a drone do it - especially if you live anywhere outside of a major metropolitan area.
What if the delivery drones have a mapping mode - so they do the mapping during the deliveries and even according to Bezos these are limited to 10 miles range from their warehouses which would be much more easier to map than swathes of cities. There is a possibility for this to succeed as well as fail, but at this point this one is more than wishful thinking - the technology at the basic level is already there.
Remember when Google brought us street view? How absurd of an ideas was photographing every street? Now we take it for granted. Is it so unbelievable that Amazon could push the boundaries the skeptics keep bringing up?
Photographing every street requires few new techniques. Mostly people didn't expect it could happen quickly or cost-effectively, or didn't expect it would be worthwhile. Arguably, it's still a neat-o factor rather than a core value.
Drone delivery requires a significant number of new techniques.
This is really just an Amazon PR stunt at this stage, but are there interesting solutions to the list of problems posed though?
Navigation: Well I suppose you could do this with an app that makes you stand in a spot where you think the drone could land, take a panorama photo which a human checks and approves. The app would also make you wait for a good enough gps fix before it 'green lights' your account for drone deliveries.
Obstacle avoidance: Maybe in 5 years time the processing power will be less energy intensive? You can't rely on maps entirely, but maybe laser scanned neighbourhoods and low resolution radar or sonar which aborts the delivery if inconsistencies are detected?
Safety: Maybe it drops down a virtual 3m 'tube' from obstruction free public airspace into the spot you send in from your app to get your landing site approved. The app would make you pick a space to land in that was suitable and you lay out a target pad. If the target pad visual looks wrong, e,g. black and white squares obscured because kid left a bike on it, then delivery is aborted. Maybe catastrophic loss of power or large downwards deceleration causes an airbag to deploy.
Legality: Some kind of peer to peer air traffic control standard is needed. Like a swarm of starlings.
Cost: Must be cheaper than a delivery driver, right? Not sure if its cheaper than a self driving car though.
It doesn't need to be cheaper than a delivery driver. The service would still be an enormous benefit to Amazon even if it operated at a loss because they'd be getting tons of business that would otherwise go to brick-and-mortar stores.
So, I get that so far this is just an Amazon publicity stunt and there isn't actually anything here to talk about.
But give them some credit. Most of these problems are solvable. You can build a drone with bumpers around the propellers and screens above and below, to solve the "sticking finger in the propellers" problem.
You can deal with the vast majority of the navigation problem by just getting enough altitude. Above a certain altitude, you don't suddenly see new unknown buildings springing up out of nowhere, and don't need to worry about navigating around power lines or anything of the sort. Just do the takeoff in a controlled area, rise up to an altitude high enough to avoid anything but a few pre-programmed skyscrapers and radio towers, and fly to your destination at that altitude.
And you can solve the landing problem by having someone sit in a control room and take over for the landing and takeoff from customer locations. Paying someone for two minutes worth of landing and takeoff is much cheaper than paying them for the full 30 minute delivery time, along with the expense of a full sized vehicle and insurance that goes along with it.
Everyone seems to worry about the liability aspect, asking "what happens when one of these falls on someone". But we already deal with liability issues with cars all the time. Heck, we deal with the issue with planes as well; planes do sometimes fall out of the sky and kill people, but you don't hear people saying that delivery by plane will never work. That's just built into the cost of delivery. Drones are much lighter, don't have tanks of gasoline, and flying in the air gives you much more open space than driving on roads. And you could probably add a failsafe that deploys a parachute if the drone loses power or otherwise start to fall.
The big problem is flight time. Batteries just don't have the energy density to support long flight times; you can build ultra-light drones that get an hour of flight time with a lot of care, but if you're building something that's actually going to carry packages, deal with wind, and so on, 30 minutes is probably the upper limit. That means that you won't be able do trips of more than about 15 minutes each way. Now, since you can go in a straight line, without traffic, that gets you further than a 15 minute car/bike delivery, but it'll still be limited to people who are fairly close to a distribution center.
Are there other types of drones other than traditional quadcopters, for example variable-pitch quadcopters [1], that can get more airspeed and thus cover more distance in 15 min?
What about using airplane-style drones (like traditional RC planes) + a parachute to land the package? And this way you could use nitro gas fuel to get better energy density (since you can use one big gas engine instead of having to use four tiny motors)
Sure, there are potentially other designs that will get somewhat higher speed, though at some point that will require enough extra power that it won't be worth it.
Airplane style drones can fly further, but navigation is more difficult, and parachutes will pose a lot more logistical problems than landing quadcopters (or, more accurately, multicopters; they used octocopters in the demo). One of the nice things about multicopters is how easy control and navigation systems are compared to something like an airplane or traditional helicopter.
The biggest improvement would be from better battery technology. Battery technology is improving, but very slowly compared to, say, transistor density.
I have quite strong opinions about this. The only way to ensure efficient, high-load drone flight is to develop a hybrid design which uses wings to generate lift during cruise. The efficiency lost to the rotors is massive.
So I think eventually we will see tilt-rotor designs, where wings take over once the drone is airborne but the rotor disk is pitched horizontal for takeoff and landing. This massively increases energy efficiency, and makes it possible to go a lot faster.
The reason this hasn't been done with larger airplanes (the only prominent example, V-22 Osprey, demonstrates the difficulty), is that combustion engines makes the design prohibitively complex. But with electric motors, it should be very doable.
Wow, that's with a greater degree of mechanical simplicity than I thought possible. But this is exactly what I was talking about. Now if you can somehow reduce propeller drag at high horizontal speeds and also accomodate a payload and/or a larger battery and also handle adverse wind conditions, you'd have something really nice.
Keep in mind, the public relations aspect of this product is as much in development as the technology itself.
Is this a PR stunt? Yes and no. On one hand it's a PR stunt in that it teases people with pie-in-the-sky products that don't exist yet.
On the other hand, this is a completely new product built on completely new technology that most people don't have any experience with. Releasing this sort of stuff is delicate because people need to be acclimated to the existence of the technology before they can be comfortable with it. This release is the very first step in that process. As the project matures, I predict they will lead the launch with a great deal of public-relations-type announcements like this.
I get that drone delivery is feasible, and in a few years COULD happen, but seriously, does this solve a problem?
If I need something like toilet paper, paper towels, nails, screws, lightbulbs, etc.. I am surrounded by places I can buy these items at in less than 30 minutes.
I admire the interest in pushing this technology forward, but I don't see the problem this really solves...
Laziness! Imagine ordering toilet paper without leaving the toilet! Hopefully Amazon will take the logical next step and have the drone drop my product with a smaller, indoor robot which can open my door via Lockitron and bring it directly TO me. Then branch out to fast food.
In seriousness, it has some utility. While I can buy toilet paper, paper towels, etc very close, it could be (for me where I live at least) considerably more difficult to find a hot new electronic item, a specific movie on Blu Ray, an unusual tool, paint in a certain shade, etc.
26 comments
[ 169 ms ] story [ 1509 ms ] threadOr are you saying that unmanned flying machines which deliver products in under 30 minutes is not noteworthy enough for people to take seriously?
“this is early, this is still years away.”[1]
Yes, 30 minute drone delivery would be big news. That Amazon is in the very early stages of an R&D project to look at the feasibility of it is not. In any case, I am staking an early bet that you'll have your Amazon products delivered by a self driving van before you'll have a drone do it - especially if you live anywhere outside of a major metropolitan area.
[1] http://techland.time.com/2013/12/01/amazon-bezos-drones/#ixz...
I wonder where a company that constantly had a fleet of drones in the air could get up to date aerial maps...
Photographing every street requires few new techniques. Mostly people didn't expect it could happen quickly or cost-effectively, or didn't expect it would be worthwhile. Arguably, it's still a neat-o factor rather than a core value.
Drone delivery requires a significant number of new techniques.
Navigation: Well I suppose you could do this with an app that makes you stand in a spot where you think the drone could land, take a panorama photo which a human checks and approves. The app would also make you wait for a good enough gps fix before it 'green lights' your account for drone deliveries.
Obstacle avoidance: Maybe in 5 years time the processing power will be less energy intensive? You can't rely on maps entirely, but maybe laser scanned neighbourhoods and low resolution radar or sonar which aborts the delivery if inconsistencies are detected?
Safety: Maybe it drops down a virtual 3m 'tube' from obstruction free public airspace into the spot you send in from your app to get your landing site approved. The app would make you pick a space to land in that was suitable and you lay out a target pad. If the target pad visual looks wrong, e,g. black and white squares obscured because kid left a bike on it, then delivery is aborted. Maybe catastrophic loss of power or large downwards deceleration causes an airbag to deploy.
Legality: Some kind of peer to peer air traffic control standard is needed. Like a swarm of starlings.
Cost: Must be cheaper than a delivery driver, right? Not sure if its cheaper than a self driving car though.
But give them some credit. Most of these problems are solvable. You can build a drone with bumpers around the propellers and screens above and below, to solve the "sticking finger in the propellers" problem.
You can deal with the vast majority of the navigation problem by just getting enough altitude. Above a certain altitude, you don't suddenly see new unknown buildings springing up out of nowhere, and don't need to worry about navigating around power lines or anything of the sort. Just do the takeoff in a controlled area, rise up to an altitude high enough to avoid anything but a few pre-programmed skyscrapers and radio towers, and fly to your destination at that altitude.
And you can solve the landing problem by having someone sit in a control room and take over for the landing and takeoff from customer locations. Paying someone for two minutes worth of landing and takeoff is much cheaper than paying them for the full 30 minute delivery time, along with the expense of a full sized vehicle and insurance that goes along with it.
Everyone seems to worry about the liability aspect, asking "what happens when one of these falls on someone". But we already deal with liability issues with cars all the time. Heck, we deal with the issue with planes as well; planes do sometimes fall out of the sky and kill people, but you don't hear people saying that delivery by plane will never work. That's just built into the cost of delivery. Drones are much lighter, don't have tanks of gasoline, and flying in the air gives you much more open space than driving on roads. And you could probably add a failsafe that deploys a parachute if the drone loses power or otherwise start to fall.
The big problem is flight time. Batteries just don't have the energy density to support long flight times; you can build ultra-light drones that get an hour of flight time with a lot of care, but if you're building something that's actually going to carry packages, deal with wind, and so on, 30 minutes is probably the upper limit. That means that you won't be able do trips of more than about 15 minutes each way. Now, since you can go in a straight line, without traffic, that gets you further than a 15 minute car/bike delivery, but it'll still be limited to people who are fairly close to a distribution center.
What about using airplane-style drones (like traditional RC planes) + a parachute to land the package? And this way you could use nitro gas fuel to get better energy density (since you can use one big gas engine instead of having to use four tiny motors)
[1] http://www.youtube.com/watch?v=Vy5Ky50eGJs
Airplane style drones can fly further, but navigation is more difficult, and parachutes will pose a lot more logistical problems than landing quadcopters (or, more accurately, multicopters; they used octocopters in the demo). One of the nice things about multicopters is how easy control and navigation systems are compared to something like an airplane or traditional helicopter.
The biggest improvement would be from better battery technology. Battery technology is improving, but very slowly compared to, say, transistor density.
So I think eventually we will see tilt-rotor designs, where wings take over once the drone is airborne but the rotor disk is pitched horizontal for takeoff and landing. This massively increases energy efficiency, and makes it possible to go a lot faster.
The reason this hasn't been done with larger airplanes (the only prominent example, V-22 Osprey, demonstrates the difficulty), is that combustion engines makes the design prohibitively complex. But with electric motors, it should be very doable.
Is this a PR stunt? Yes and no. On one hand it's a PR stunt in that it teases people with pie-in-the-sky products that don't exist yet.
On the other hand, this is a completely new product built on completely new technology that most people don't have any experience with. Releasing this sort of stuff is delicate because people need to be acclimated to the existence of the technology before they can be comfortable with it. This release is the very first step in that process. As the project matures, I predict they will lead the launch with a great deal of public-relations-type announcements like this.
If I need something like toilet paper, paper towels, nails, screws, lightbulbs, etc.. I am surrounded by places I can buy these items at in less than 30 minutes.
I admire the interest in pushing this technology forward, but I don't see the problem this really solves...
In seriousness, it has some utility. While I can buy toilet paper, paper towels, etc very close, it could be (for me where I live at least) considerably more difficult to find a hot new electronic item, a specific movie on Blu Ray, an unusual tool, paint in a certain shade, etc.
Faster consumption of more stuff, the goal of every company.