Quadcopters
Quadcopters are devices which fly by spinning their 4 helicopter-like blades, normally located on the 4 corners of the quadcopter. Beyond that, the definition is fairly open -- there's no real definition of payload or electronics beyond having a big enough battery to fly. Quadcopters can be remote-controlled, or they can be automated drones. The Khan Academy has a great video covering all the basics.Billboards
Most people who use computers are familiar with picture elements, aka pixels. Most of us also know that an HDTV has 1920 * 1080 = 2,073,600 pixels, which is almost 2 megapixels. If each quadcopter were 1 pixel, if each one provided a single 3-color LED of light to the billboard, you'd need 2 MILLION quadcopters to make an HDTV billboard. If you could make them for $50 each, the billboard would cost more than $100 million.
What about stadium big screens? Huskytron, the large screen in the University of Washington Husky Stadium billboard was built by Sony for $3 million, and has 255,216 pixels across its 16 feet x 9 feet. This puts its resolution at about 640 x 400, and about 40 pixels per foot, or about 3 per inch. The cost per pixel is ($3,000,000/255,615) $11.75. With $50 quadcopters, a display of that size would cost about $13 million.
Mini Quadcopters
So where do we find quadcopters for $50?
These cute little 4"x4"x3" Walkera QR Ladybird Mini Quadcopters are on Amazon for $55 each. I would assume a discount when purchasing in bulk, so even with custom electronics, something this size should be sourceable for under $50 each.
If we wanted to be slightly more conventional, we could use RC helicopters. This one runs just $20, so it could halve the cost of the billboard.
Quadcopters of a feather
Since these quadcopters (or RC Helicopters) would be flying in tight formation, there are aerodynamic concerns about how closely they could fly. Let's assume these quadcopters could fly if they were spaced 1 foot apart. Our HDTV billboard would require an 1100-ft tall, 2000-ft wide airspace -- it would be 3 or 4 city blocks wide, and would almost certainly require either FAA approval or the approval of landowners. If we hung the HDTV billboard 100 feet off the surface of Elliott Bay, facing Queen Anne hill, and the quadcopters were "stacked" in vertical columns, the top pixels would be 1100 feet from the eyeballs of people on the ground, and about 650 feet from those on top of Queen Anne hill.The "low-definition" billboard would only require a 500-ft tall, 650-ft wide airspace, which is only 2 or 2.5 city blocks wide. The FAA would still want to be involved. If we hung the low-def billboard in the same spot, its top pixels would only be 600 feet from those on the ground, and 144 feet from those on the hill.
At this scope, the billboard may as well wrap around part of the city, and wrap overhead as well -- so the quadcopters would sit in a kind-of partial dome formation.
Brightness
What if we hung this flying billboard 1 mile away -- how bright would the LEDs have to be for us to see them? Well, apparently LEDs exist which can do exactly this. I'm unsure if 3-color LEDs exist which are visible from 1 mile away, but the worst-case option here is to find bright, clear LEDs and paint them. Since this device runs on a single CR2032, the battery and LED should add very little to the weight of the quadcopter. These could be easily controlled by an onboard Arduino or other processor, and control of the entire screen could easily be transmitted wirelessly.
Control
Here's a cool video of 49 synchronized drones from the Ars Electronica Futurelab in Linz, Austria:Now we simply need to multiply that by 50,000. :)
Orchestrating the synchronized flying of 260,000 devices is no small feat. Only the largest server farms currently have that many devices. It would be resource-prohibitive to use remote computers for flight control for so many devices. Instead, these devices could handle the basics of flying themselves, and be given limited direction information by an overlord drone quadcopter. As the Khan Academy video showed, we can have computers control these devices with a few million lines of C++ code. And many embedded processors can now handle a few megabytes of code.
Instead of putting sensors around the room, we could put sensors on each display drone quadcopter. Laser pointers and photoresistors would be a "quick and dirty" way for them to easily form a grid. Yet they would still need an overlord drone to bring the video input, parse, and orchestrate it to the display drones. And there are limits to how many clients can connect to a Wifi access point -- Cisco recommends not more than 24, and other sites tend to agree, though apparently an AP can handle 2048 different MAC addresses. So even if we could max out each AP, we would still need 125 APs.
These APs would of course be mounted to overlord drone quadcopters. Following Cisco's recommendation would require more than 10,000 APs, which would require another 445 superlord drones to manage the overlord drones, 19 ultralord drones to manage the superlord drones, and 1 motherlord drone to manage the ultralords.
Additionally, most APs are designed with a 300-foot range. If an overlord drone quadcopter were near the center of the grid, I'm not sure how many quadcopters would be in range, but not more than ~300 in any straight direction, so maybe 1500 devices would be in range of an AP. Those APs on the edges would reach fewer clients.
It's difficult to speculate further on these details without being closer to development. There are many problems to solve here, but there are more solutions than problems.
Battery and Power
A point I tend to neglect when discussing quadcopters is flight time. The Ladybug quadcopter I linked earlier has a 3.7V 240mAh battery, which gives it 6-8 minutes of flight time. The battery holds almost 1/6 the energy of the 1300 mAh battery in my phone, and weighs about 1/10 as much. The entire Ladybug quadcopter weighs as much as my phone battery, so adding another battery to the device may not work so well.At 6-8 minutes of flight time, the battery would require 8-10 recharges per hour. This means the device is using 1800-2400 mAh per hour to remain airborne. This doesn't include power for the processor or LED, and it doesn't include their extra weight. For the full fleet of 260,000 devices, this is between 4 and 6 MWh per hour. In 24 hours, these batteries would have to be recharged between 180 and 250 times. This very detailed blog post explains in too much detail that Lithium batteries (LiPo, Li-ion, LiFePO4, etc) only last about 300-500 charge cycles because the lithium inside slowly changes forms from battery to crystal. Thus, we could expect to replace these batteries after about 48 hours of billboard operation, for a price of about $1.5 million.
The most elegant solution to this problem is wireless charging -- if there were a method of directly transmitting electricity through the air, then flight time would be much longer. Sadly, wireless charging technology isn't yet developed enough to be viable. Maybe it soon will be. Another solution is solar panels, but a solar panel which outputs enough to power the Ladybug quadcopter would again weigh as much as the entire quadcopter.
Something of note is that the CR2032 button battery mentioned above is a 3.0V 225mAh battery, almost as powerful as the Ladybug's battery. I wonder about the feasibility of using 1 or 2 CR2032 batteries in this device instead of these more expensive batteries.
Conclusion
So yes, quadcopter billboards are technically feasible. The biggest hurdles to overcome are:-Creating the flight management software, which is apparently being solved by engineering grad students (feel free to hire some).
-Networking 260,000 devices with wifi technology
-Getting 5-6 MWh into the air to charge these 260,000 devices.
-Battery life expectancy.
