Inside Google’s Drone-Delivery Program
theatlantic.com
theatlantic.com
But the use case that this article touched on, which to me was a "ah ha! we could do that today" is the case where the sending and receiving parties are more fixed (not just any random person's door), where they could actually build a drone-pad (like a heli-pad, but much more basic), where you could program in all of the specific obstacles on the route (exactly where, if there are electrical lines, what flight pattern to take, etc), and where you could train both the sender and receiver, and cord off the landing area (so you'd never run into the random guy or dog tries to grab the drone problem)
examples:
* delivery of medical supplies/medicine to a facility in a very remote area where it would not make sense to stock the items there permanently.
* very expensive restaurant at a ski resort in the middle of Colorado that gets a drone delivery of fresh seafood from a CA pier every day (an existing business transaction they do anyway that could be done faster, more efficiently than a human operated aircraft)
* the delivery of anything else to a ski resort (you have very wealthy people wanting things in a very remote location)
* something on top of a cruise ship, for medical items or whatever other thing a passenger really needed while at sea.
* something at the Ebola treatment areas right now in Africa.
Something like this would be incredible for allowing access to specialized medicines at drastically lower costs. There were many locations that they were unable to get guaranteed couriers, and hence couldn't offer their treatment at any price...
I wonder if it's possible to design a drop chute instead of a drop zone container so that the drone never needs to do the winch thing; the drone drops the package into the chute at it's current hover height and takes off while the chute takes care of slowing the package down to the ground.
In any case, this is an exciting area ripe for experimentation and bold ideas.
Skip hover, we don't need 10cm accuracy yet.
What's the cheapest way I can deliver one or multiple payloads from a fixed wing aircraft at 5000 feet to a 100x100 ft 'package' zone with 99.9% accuracy? (in any weather condition).
we grow quickly by letting people with big backyards sign up to be the delivery zone, then manage the last miles on their own.
Some companies that have tried include:
BufferBox (YCombinator backed, bought by Google)
business.financialpost.com/2012/11/30/google-snaps-up-waterloo-startup-bufferbox/
SwapBox
DHL and others have similar services:
http://images.computerwoche.de/images/computerwoche/bdb/1841...
In the future it will be common for homes to have a drop point that accepts larger parcels than mail. It will be a feature that people are going to look for when renting/buying homes. With that in place and a good standard drone delivery will also become more practical for the last mile.
Ideally someone would step up and develop a standard similar to what happened to shipping containers. That would help the adoption tremendously.
But people aren't doing that now. Why not?
By having a standard UPS/.. could be notified that you have a "A1" container available. Any parcel that fits in that container can be delivered without arranging a meeting time. It could be done today in rural households but in cities, room would need to be made for these new containers.
I've heard of quite a few S&R stories where they knew where the person was but couldn't safely get there until daylight. It seems like being able to drop off a care package in such situations would be useful.
In some situations like harsh cold snowy weather a drone is unlikely to be of much use, but dropping a couple of liters of water to someone stuck overnight in the desert could be the difference between pulling in a dangerously dehydrated lost hiker in the morning versus picking up someone who just needs a lift back to civilization.
Of course, drones (even sans delivery ability) seem like a good fit for S&R anyway -- relatively cheap (compared to manned helicopter searches) and safe way to cover last known areas of missing folks to try to locate them from the air.
It has the added advantage of being high-importance but low-cash-value, reducing any incentive from (hypothetical) drone-nappers and making the experience a positive one for the general public in the area of operation.
https://en.wikipedia.org/wiki/Tail-sitter
This means it's much faster and more efficient in flight than your standard quadcopter design.
Also, Google's full official promotional video, the BBC clip seems to be taken from this.
The QuadShot seemed a bit difficult to control. The automatic control and/or the larger size seemed to help the Google drone to be more stable.
It is an interesting design choice. I hope to see more updates soon!
I have a QuadShot and it's not too difficult to control. I think with a good control system the airframe could work pretty well. The QuadShot is quite lightweight, has all its mass centered, and has a pretty large wing cross-section which makes it difficult to fly in the wind though. I also don't feel that the QuadShot's firmware was ever quite nailed down completely - I feel that there's still a lot more stability to be gained by tweaking PIDs and improving the control system, or by adding additional inputs like optical flow.
The Google device probably benefits a lot from being bigger as it's difficult to substitute for inertia.
The k-max helicopter was used for a lot of unmanned cargo supply missions.
The only problem with tail-sitters is the pilot (and possibly the payload).
Actually, a tilt-rotor can be manipulated to make the aircraft go where you want it to. The same idea explains how helicopters make their way across the landscape.
"In all the testing, Roy had never seen one of his drones deliver a package. He was always at the takeoff point, watching debugging information scroll up the screen, and anxiously waiting to see what would happen. “Sergey [Brin] has been bugging me, asking, ‘What is it like? Is it actually a nice experience to get this?’ and I’m like, ‘Dude, I don’t know. I’m looking at the screen,’” Roy told me."
Google and Amazon are both great companies. But at Amazon the drone program would start with a description of what the customer experience is when they receive a drone delivery and you'd work backwards to the technology solution. At Google the technology precedes the customer experience.
"Another obvious idea is to simply land the craft, drop the package, and then take off again. To test the premise, they brought in some of Google’s user experience researchers who queried people about how they might react to such a delivery. What they found was that individuals could not be stopped from trying to reach for their packages, even if they were told that the rotors on the vehicle were dangerous, which they are."
I have worked for neither company but I use products and services from both daily and I have to say I disagree. And remember the Prime Air demo video that showed a drone landing in a yard? And when your dog thinks it's a great new toy and runs up to it? Or your kid?... I have to say they didn't really think that experience through before taking it to the press.
While the hardware is a significant part of the problem, they seem largely agnostic about which flying machine might ultimately serve their needs best. The real challenges, Teller and Roy insist, come in the design of the rest of the system like, for example, the delivery mechanism.
For example, the reason they do the winch down thing, instead of just landing, is so that customers are not tempted to approach the drone and get hit by the rotors.
"Don't make your problem the customer's problem!"
There are often many technical solutions to create a particular customer experience. Jeff never accepts false dichotomies. If the particular technical solution being considered fails to produce a delightful customer experience then you better start thinking about other solutions. Many of the projects on which I worked turned out much better when Jeff challenged us break out of our little technical boxes and consider alternate solutions that preserved the customer experience.
I can sense that my answer will be frustrating to some who don't like being told that they aren't thinking big enough / creatively enough / bold enough / outside the box enough. Believe me, I felt that way too when working on Jeff projects. I left his office in frustration many times only to realize several hours or days later that I would have been better served by focusing on creative problem solving instead of trying to convince Jeff that X was not possible.
I have immense respect for Jeff. I also have increased respect for Sergy because he appears to be asking the right types of questions at Google.
Speaking of false dichotomies.
I think Google, Amazon and even Microsoft are hedging similar bets against the limits of technology. After all, what's impossible today may be trivial tomorrow.
CERN built capacitive multi-touch displays in the 70s. Mitsubishi, Bell, Microsoft, IBM and a bunch of other companies were involved in the R&D from there.
What Apple are good at is integrating and commercializing at the exact point where the economies of scale are viable, adding a layer of accessibility for users (eg. inventing the gestures, but not the touch technology itself).
The anecdote about how Apple tried to launch the iPad before the iPhone is proof that they didn't catch-on to how the core touch tech scaled (every inch of touch surface would increase production cost non-linearly) until much later.
"work backwards to the technology" is a quote from Steve Jobs:
> One of the things I’ve always found is that you’ve got to start with the customer experience and work backwards to the technology. — http://youtu.be/GnO7D5UaDig?t=52m16s
That a quote like that should be attributed to Amazon shows how much Jeff is trying to emulate Steve.
Initial announcement https://news.ycombinator.com/item?id=6830547
Don't believe the hype https://news.ycombinator.com/item?id=6833223
Delivery drones are nonesense https://news.ycombinator.com/item?id=7005702
Is Amazon drone delivery real? https://news.ycombinator.com/item?id=6834561
Also worth noting is that Australia has a "UAV Outback Challenge" [1], where UAVs compete to carry out a simulated rescue of a tourist lost in the outback. This September the challenge is in Kingaroy, Queensland. (Maybe that's why Google is testing up that way??) Of further interest, there's at least one open source team competing in the UAV challenge [2].
[1] http://www.uavoutbackchallenge.com.au/
[2] http://www.canberrauav.com/
------
Edit: A bit more detail in the local news: http://www.smh.com.au/digital-life/digital-life-news/google-...
[1] That being somewhat more than a http://en.wikipedia.org/wiki/Sikorsky_S-64_Skycrane skycrane can deliver, and with better failover characteristics.
When the tech and price scale are right for this application, you'll definitely see it as it has been discussed and is actively being worked on.
From http://en.wikipedia.org/wiki/Pyrocumulonimbus_cloud:
"On the 18 of January 2003, a supercell thunderstorm formed from a pyrocumulonimbus cloud associated with a severe wildfire, during the 2003 Canberra bushfires in Canberra, Australia. The supercell resulted in a huge fire tornado, rated at EF3 on the fujita scale, the first confirmed violent fire tornado. The tornado and associated fire killed 4 people and injured 492."
That is an interesting link. Reading about the Australian fires an interesting question comes to mind which is this, "If we engaged these fires at the outset with a constant suppression/extinguishing scheme, would they still get to the size and ferocity that they do today?" that is something I don't know. CalFire jumps on fires pretty quickly to keep them under control, especially near structures, and those fires then evolve slowly. Unlike say the Yellowstone fire a decade or so ago which was left to burn 'naturally' and it got quite large.
I grew up in the Australian bush, and it, like the Californian Eucalypt forests, is prone to burn at the slightest provocation. Add dry, windy conditions and it's a recipe for disaster.
Small electric stuff will be limited by charging infrastructure and plain availability; it'd be fun to see a future where various commercial swarms exist and they're taken over for disaster relief.
"Hey, where's my burrito?"
"Sorry sir, all our BurritoCopters[tm] are attending the San Bernardino Valley Fire at the moment, per civic ordinance K-34. Your burrito's right here if you want to collect."
Or, to push this idea further, maybe we could repurpose artillery and just shoot water buckets from long range? I wonder if this or the missiles would have lower fuel cost than Skycrane or drone operations.
You also use them for single-use things (missiles) because the value in making a thing blow up at a distance is a lot higher (a $5 million dollar tank, for example) then the rocket itself.
But they're terrible for when you're into purely shifting mass. Air-breathing propulsion with aerofoils is absurdly better. It's why hypersonic flight and things like the Skylon are a big deal.
That said, I could imagine a world where a FedEx drone is permitted to provide last meter service from the roof of a truck, taking the package from the street to the doorstep. Additionally, widespread use of drones may reshape where core resources are stored and allocated. In Manhattan for instance, maybe lightweight warehouses are placed on rooftops for quick deliver of common products within a few minutes of ordering.
I'm now imagining a shipping container that get floated from China to $city stacked up top on a regular container ship, then without even unloading it from the ship, the roof slides back and the walls collapse, and a flock of disposable single-use drones launches out - delivering random manufactured crap directly to the purchaser bypassing the expensive first-world logistics chain...
When you mentioned the shipping container, I was thinking about a heavy lifting helicopter that plucks an entire shipping container off the cargo ship and drops it on an 18 wheeler already on a highway (or on a train already in motion...). Bypassing the bottleneck at a port could save a lot of time and money, too.
The mobile app would then tell you here are 50 drones with a 10 mile radius - & here is the list of items that are on them - so forget same-day delivery how about same-hour delivery?
If Tuesday was chicken curry night and the cost reflected avoiding the variance and random delivery costs that drive up prices for other food services, I'd buy every week.
Driver plus someone to hand over the food or leave it at the door.
I'm curious what kind of running time these have though and distance they can travel. Would be great if they could just have 2-3 go for a dozen roundtrips each in circumstances like Katrina.
That is just their plan for getting regulatory approval: if you are against UAVs, you are against delivering life-saving medicine to disaster victims.
My mailbox is not exactly a vault, but it's enough of a deterrent to stop privacy violations and petty theft.
Good allusion, and I didn't even like that damn book!
https://www.youtube.com/watch?feature=player_embedded&v=kDay...
https://www.youtube.com/watch?v=pi4jZKqZ8v8
Delivery was of a salami sandwich (we later cleared it out with the police).
I even gave Ryan Hickman at Google the software and hardware specs, when he was trying to fix up the Cellbots project in 2011. We were at maker faire 2011, which was funny because their android 'bot didn't work and ours did... Never heard much since, though.
Sorry for bragging, but it's a pet peeve I got. Anyway, the source and schematics are at http://obex.parallax.com/object/116 the Android side software can be had if you email me at mkb@robots-everywhere.com
We still use this to do autonomous deliveries! https://www.youtube.com/watch?v=urs68vf7ZFY
By an extremely liberal interpretation of "this", perhaps.
What's your pet peeve? That you did something vaguely similar a few years ago? Someone did it before you, too.
I'd like to know who else has built and successfully trialled tail-sitting autonomous flying vehicles. This stuff is awesome!
http://dasl.mem.drexel.edu/alumni/bGreen/www.pages.drexel.ed...
https://www.microstrain.com/sites/default/files/Autonomous%2...
http://uav.ae.gatech.edu/wiki/uavrf/movies
Look for GTEdge transition to or from hover in the video descriptions. (We probably weren't the first either...)
Could be publication delay or a mistake. Same era, so the time had come for that work.
Also, thanks for downvoting a post with link to source and schematics, whoever you are.
I know I'm burning a lot of points on this one, but the fact is, I wasn't any more of an egomaniacal ass than the guy I talked to who tried to frankly pretend my work didn't exist.
And, this site working how it does, most of the value is in the comments -- why downvote something with (maybe tangentially) related links to open source software/diagrams that other people can use? This is HACKER news, correct?
(Now watch me get downvoted to oblivion just because I am standing up for what I think and backing it up as rationally as I can manage)
Well, the "this" operator is overloaded, just like in C++.
We used a G1 dev phone to fly a 1.5 meter wingspan airplane. We used only sensors in the phone, camera, gps, etc.
Honestly it didn't work too well - we needed either a much better model of our airframe, an airspeed (not groundspeed) sensor or an airframe that could power through 10-15 mph winds.
In the end however, we demoed waypoint navigation and imaging the waypoint. You text it lat/long and it texts back a picture.
We did it as a proof of concept (for an interested govt agency), modern smartphones are neat, but bespoke, real-time hw, with navigation grade sensors are probably an order of magnitude more effective (though an order of magnitude more expensive).
Working with android to do robotics was.. interesting. We had to do all sorts of hacking to make the phone behave a little as possible like a phone and more like a real-time controller. What do you do when someone calls the phone while you're flying an airplane? How do you keep the phone application from taking over and exiting out of your navigation app? While the airplane is 1000 ft in the air?
What I ended up using for the ground/water stuff was sending serial commands out of the headphone jack... surprisingly low latency, especially compared to bluetooth, and it was only a few bytes at a time so low baudrate was fine.
http://robots-everywhere.com/re_wiki/index.php?n=Main.AudioS... Here's the source and schems if anyone wants them -- the idea here was to make the hardware as light as possible!
Do you want Google acknowledge that you did it first?
That this drone project may not have been possible without your earlier efforts?
Does it use maps&gps/vision (yes)/lidar? (I assume the green images are taken from the imaging sensor on the controller).