Surely they can’t be using GPS for positions just a few feet apart.
Also, how quickly can they respond to wind and such?
In: 699
Each of them has a little onboard computer that is able to adjust for things like the breezes and what not.
So they all fly a preprogrammed course, with just a little bit of individual individual judgment given to each machine based on variation and things like wind.
And realistically, a large enough wind would screw that up. Which is why these drone shows tend to occur in places like football stadiums where there is some degree of protection.
Computer programming. They aren’t being live-piloted. The drones are controlled by a program that is sending them signals where to be, what color lights to show, etc
Each drone has a number of sensors on it. Accelerometers that determine how fast it moves in 3D and which way gravity is, and electronic gyroscopes to determine which direction it’s turning, for example. These two sensors are cheap enough to be everywhere and fast enough that, with proper programming, they can keep a drone hovering stationary in any condition the propellers are strong enough for (edit: by sensing the movement caused by wind or being pushed, and then propelling the drone in the opposite direction until the movement has been cancelled). They can also be combined to map where something is, accurate to within centimeters if calibrated correctly, just by keeping track of how they move. However, that calibration is tricky and sometimes drifts, so it probably isn’t the only tool.
A system of lights on the drones can be used in proper lighting (whether that’s bright enough or dark enough depends on the system). This could use a camera from the controller’s perspective and special lights or reflectors on the drones to easily determine where and how far away each drone is, and update the drone’s control with constantly corrected calibrations. It would also tell the drones where they need to be for the next part of the show and how to get there without hitting each other.
There are also probably other methods, and the more methods used the better/more accurate/safer the display is, but even with just these two most drone displays would be completely successful.
At least some of them use real time kinematic GPS. This involves having one or more base stations no the ground that provide much more precise and faster positioning.
[https://en.wikipedia.org/wiki/Real-time_kinematic_positioning](https://en.wikipedia.org/wiki/Real-time_kinematic_positioning)
>Surely they can’t be using GPS for positions just a few feet apart.
* Why not?
>[Accuracy of differential-grade GNSS units varies depending upon the type of differential correction applied and the quality of the GNSS receiver and antenna (type, quality, and the number of satellite and frequencies that can be received), with external antennas typically providing the best results.
>
>
>
>All Differential-grade GPS receivers have a horizontal positional accuracy of less than 1 meter. Most new GPS receivers with differential corrections from SBAS such as WAAS and low level OMNISTAR subscriptions or from GBAS such as beacons typically have accuracies from 0.3 to 1.0 meter, depending on the quality of the receiver. Higher-level OMNISTAR service or Trimble’s H-Star service improves the accuracy to 5 – 30 cm. Currently, the highest quality differential GPS receivers available are dual frequency units that utilize both GPS and GLONASS satellites.These coupled with a very accurate differential correction subscription will give the best differentially corrected position possible. Vertical accuracies for these GPS units are 2 – 3 times that of the horizontal accuracy, and should be used only for informational purposes.]([https://water.usgs.gov/osw/gps/](https://water.usgs.gov/osw/gps/))
GPS alone can’t be used. The reason is that the US GPS signal is intentionally degraded for non military applications. Another reason is atmospheric conditions that reduce precision.
The most common solution is using differential GPS. One GPS receiver is placed in a known position at ground level. This devices calculates the error in the GPS signal and sends it to the drone. The drone use the error correction information and the satellite signal to reach centimeter level accuracy. This is the strategy used in the 2020 show involving 2198 drones.
Another solution is to have only one drone with GPS. The other drones use an ad-hoc network to find their position relative to the first drone. Even if the first drone is not positioned precisely, the other drones will be displaced by the same distance.
The reaction to wind is quick. The drones have internal accelerometers to instantaneously detect if they have been moved from their position. The corrections happes in less than one second.
EDIT: for shows involving a limited number of drones, GPS alone can be sufficient. If the drones are not too spread apart, their receivers will all be wrong by the same amount. The swarm will be displaced some meters away from the target position, but the relative position of the drones will be correct.
Sub centimetre resolution positioning systems, with base stations providing a local reference locating beacon
All these responses focusing on just GPS are missing one obvious way to do this: if the drones know how far they are away from other drones to a reliable degree, then you can you can use the rough GPS position of the entire formation.
You can do that with some combination of radar, laser, or ultrasonic range finding.
Maybe I’m just plain dumb but couldn’t they just be preprogrammed to do a specific up down sideways routine and it’s all just programmed from where they take off. My programming skills are limited and the only robotics stuff I did was FLL over a decade ago but I’d say instead of over completing it wouldn’t it make sense just to have each drone have an approved route and not have them chatting with each other?
I think they use WLAN like the [FLARM collision avoiding system of gliders and small GA planes](https://flarm.com/).
Latest Answers