How is this done?

[A64Pilot]

The aerial light shows above the house are apparently 300 Drones, so how do they position so exactly? GPS is good, but not that good. I’m assuming the position compared to each other by some kind of onboard sensors, maybe one is positioned via GPS and the others take position off of it?

 

[Mark89114]

Good Question......below is not a good technical answer.....but some clues.  I would think they have to cue off of one primary drone to follow.    

 

Swarm technology enables multiple drones to communicate with each other like bees, working together towards a shared end goal rather than having each drone’s flight path individually prescribed. Using GPS or radio frequencies to position themselves, drones can wirelessly chat to each other to ensure they maintain their spacing and avoid collisions. They can also determine which of the drones is best suited for each job required to bring the show to life.

 

[N201MKTurbo]

If I were to do it, I would write the software to define lines and arcs in 3D space. I would then break them up into line segments and make a voxel at every vertex. The voxels could be viewed in a point cloud viewer to verify it was working the way you like. I would then use 3D matrix transformations to animate the point cloud. Changing shapes would be difficult to keep them from crashing into each other. That's where the swarm technology would help. 

While doing a shape, I would have a way to communicate with each drone individually, continuously updating their GPS position setpoint. And light color and intensity.

They could all use GPS position. Any GPS errors would affect all drones the same and just cause the whole swarm to follow the error.

[EricJ]

2 hours ago, N201MKTurbo said:

They could all use GPS position. Any GPS errors would affect all drones the same and just cause the whole swarm to follow the error.

And differential GPS, even using a temporary ground transmitter, makes a big difference as well.

There are also techniques for range finding between radios that are pretty accurate for line-of-sight link distance, so if they use that between elements they'll have additional inputs for feedback on the geometry.

[N201MKTurbo]

1 hour ago, EricJ said:

And differential GPS, even using a temporary ground transmitter, makes a big difference as well.

There are also techniques for range finding between radios that are pretty accurate for line-of-sight link distance, so if they use that between elements they'll have additional inputs for feedback on the geometry.

Yes, but all differential correction would do is remove the errors. Who cares if the whole formation is offset a foot or so one way or the other?

[N201MKTurbo]

Can you imagine charging the batteries on 300 drones?

[N201MKTurbo]

https://www.chetu.com/aviation/drones.php?keyword=drone swarm software&gclid=EAIaIQobChMIkPS3rJLy-gIVMRWtBh1DHAc_EAAYASAAEgJfPPD_BwE

https://droneshowsoftware.com/

https://verge.aero/everything-about-drone-light-shows

The site above said that a one time show like the one in the video costs about $20000

[jaylw314]

8 hours ago, A64Pilot said:

The aerial light shows above the house are apparently 300 Drones, so how do they position so exactly? GPS is good, but not that good. I’m assuming the position compared to each other by some kind of onboard sensors, maybe one is positioned via GPS and the others take position off of it?

I recall seeing systems that positions each drone relative to its nearest neighbors via formation lights and sensors, without the benefit of GPS.  The idea was to make a system that could be used indoors, like gymnasiums or indoor arenas.

[Marauder]

Don’t know where you guys get your information from. But I know for a fact that these drones are in reality small crafts piloted by little green people who practiced the choreography for months. Just like the Blue Angels.



You might have caught one of their earlier shows in “batteries not included”.






Sent from my iPhone using Tapatalk Pro

[EricJ]

36 minutes ago, jaylw314 said:

I recall seeing systems that positions each drone relative to its nearest neighbors via formation lights and sensors, without the benefit of GPS.  The idea was to make a system that could be used indoors, like gymnasiums or indoor arenas.

I don't know how mature they are yet, but several wireless standards (e.g., BT, WiFi) had looked at some efforts for relative location using ranging and beamforming, the motivation being finding items in warehouses, etc., etc.   Line-of-sight ranging is not too difficult, and with multiple devices in the network arranged in 3D that alone could provide a lot of information relative position.

I don't know whether that's what's being done or not, but it's a reasonable methodology that could be used.   Regardless, it's not too mysterious.

[A64Pilot]

I’ve worked with GPS significantly and especially before WAAS, particularly when SA was enabled it took differential GPS to be able to say drive a tractor plowing a field, or any kind of decent accuracy.

But differtial isn’t cheap nor tiny and light, or wasn’t anyway, but I think in order for swarm techniques to be very effective, they are positioning themselves off a lead aircraft and of course in all three axis, which if true is very impressive to me.

By the way a GPS accuracy of even 1 meter wouldn’t shift the whole formatation by one meter but all vehicles could be off by one meter in any or all three axis which would blow the formation, then there is the issue of GPS accuracy in altitude isn’t nearly as accurate, so I’m convinced they are somehow positioning themselves off neighboring aircraft.  However they are doing it, it sure seems to be precise

[N201MKTurbo]

13 minutes ago, A64Pilot said:

I’ve worked with GPS significantly and especially before WAAS, particularly when SA was enabled it took differential GPS to be able to say drive a tractor plowing a field, or any kind of decent accuracy.

But differtial isn’t cheap nor tiny and light, or wasn’t anyway, but I think in order for swarm techniques to be very effective, they are positioning themselves off a lead aircraft and of course in all three axis, which if true is very impressive to me.

By the way a GPS accuracy of even 1 meter wouldn’t shift the whole formatation by one meter but all vehicles could be off by one meter in any or all three axis which would blow the formation, then there is the issue of GPS accuracy in altitude isn’t nearly as accurate, so I’m convinced they are somehow positioning themselves off neighboring aircraft.  However they are doing it, it sure seems to be precise

You are incorrect. They are all receiving the same GPS signals. any inaccuracies would apply the same to all of them. They wouldn't have random errors that would apply to them individually. GPS works by measuring the time delay between the satellites and the receivers. The errors come from the reported positions of the satellites and their internal time references. All the receivers would be subject to the same errors, so they would all be calculating their positions with the same errors.

All differential GPS did was receive the GPS signals from the satellites at the fixed location and calculate its position, but it knew where it actually was and transmitted the error correction to the differential receivers. This required that both the receiver at the fixed location and the mobile receiver calculated the same errors, or the correction wouldn't work.

[N201MKTurbo]

If you read the last article I referenced above, it gives a lot of detail about how it works. There is no mention of any communication between the drones, but it does appear that they do use differential GPS. I also found it interesting that the drones don't know which drone they are until they are placed in the take off field. They report their GPS position back to the ground station and the ground station tells them which drone they are. They all have the flight path for all the drones in their memory, so all they need to be told is which one they are. It seems that after the show starts they are all autonomous.

In contrast, Verge Aero drones are much brighter and hold their position precisely by supplementing GPS with additional positional data sent from the ground station. This enables more creative flexibility and allows the same effect to be achieved with fewer drones. We can now precisely draw lines in 3D with drones.

 

[EricJ]

21 minutes ago, N201MKTurbo said:

If you read the last article I referenced above, it gives a lot of detail about how it works. There is no mention of any communication between the drones, but it does appear that they do use differential GPS. I also found it interesting that the drones don't know which drone they are until they are placed in the take off field. They report their GPS position back to the ground station and the ground station tells them which drone they are. They all have the flight path for all the drones in their memory, so all they need to be told is which one they are. It seems that after the show starts they are all autonomous.

In contrast, Verge Aero drones are much brighter and hold their position precisely by supplementing GPS with additional positional data sent from the ground station. This enables more creative flexibility and allows the same effect to be achieved with fewer drones. We can now precisely draw lines in 3D with drones.

 

There's a reasonable chance that multipath reflections to each unit will be different, so the errors at each may decorrelate due to the multipath energy.

It is true that the ranging and position locating features in some modern link systems could be exploited with a system using a number of reasonably widely separated ground stations that coordinate to correct positions of various units.   I'm not sure how much accuracy that'd provide, but it'd be something.    Or even just creating a localized gps-like system with several carefully-surveyed ground stations transmitting ranging information could do it, too.

Anyway, there are multiple ways to get there from here.    None are trivial, but apparently at least one works well and is practical to implement. 

[N201MKTurbo]

I have a few friends with drones. They have a fly to home feature where they will return to their takeoff position if the battery gets too low or if the radio link goes down. That thing will land within a few centimeters of where it took off. It does this with just GPS. That seems plenty accurate to fly these shows. 
 

And I know it doesn’t show in my byline, but I do have a commercial remote pilots license. Even though I’ve never flown a drone.

BTW, you do t want it to use the return home feature if you are flying it from a boat.

[Pinecone]

6 hours ago, A64Pilot said:

I’ve worked with GPS significantly and especially before WAAS, particularly when SA was enabled it took differential GPS to be able to say drive a tractor plowing a field, or any kind of decent accuracy.

Because the error is in the position in relationship to the ground.

If you had two tractors, they could plow in formation using GPS, but both with not be precise to the actual ground.

 

[A64Pilot]

42 minutes ago, Pinecone said:

Because the error is in the position in relationship to the ground.

If you had two tractors, they could plow in formation using GPS, but both with not be precise to the actual ground.

 

No, if you took a Gps and put it on the ground in a fixed spot and had it record positions, it would record a circle of dots, the circle would represent of course positional data, as different GPS’s don’t take positional data at the same precise instant, they will record different locations, if you could precisely synchronize them , then they would agree I guess. The tractors would weave about and not together. The inaccuracy isn’t constant, it’s random, with SA back in the day it added 50 meters of error horizontally and 100 meters vertically, without WAAS I think best accuracy possible was 10 meters. Neither would steer a tractor, that needs to be within inches, so enter Differential GPS which takes you from the average 15 meters to roughly 2 cm, good enough for the tractor to be more accurate than any farmer, this increased crop yields significantly because the GPS tractor being more accurate would plow more rows in the same acreage.

My first personal Handheld an Apollo Precedes had a survey mode, it would continue to take and record measurements until you stopped it and would take the average of those measurements and give you a very precise location as it averaged out the positions and could get you well within 10 meters, but survey mode of course was only useful if stationary.

A model AH-64 had no GPS and I would get a precise location data for us to align our INS and do a position update before a mission. Initially we used maps to get location, but in the desert a map isn’t much good as the terrain changes

Anyway without differential GPS I don’t think those drones could be so precise, but perhaps they have differential GPS now, first GPS’s weren’t small. But for years there have been GPS watches, so who knows? It seems differential GPS is likely?

I just wondered if anyone knew.

[A64Pilot]

3 hours ago, N201MKTurbo said:

I have a few friends with drones. They have a fly to home feature where they will return to their takeoff position if the battery gets too low or if the radio link goes down. That thing will land within a few centimeters of where it took off. It does this with just GPS. That seems plenty accurate to fly these shows. 
 

And I know it doesn’t show in my byline, but I do have a commercial remote pilots license. Even though I’ve never flown a drone.

BTW, you do t want it to use the return home feature if you are flying it from a boat.

I’ve had an inexpensive one for awhile, it will even follow me, or more likely the transmitter of course and circle you if you stop, camera pointed at you. I bought it to take pictures of us sailing, but never did, probably would have put it in the water landing the thing.

[EricJ]

2 hours ago, Pinecone said:

Because the error is in the position in relationship to the ground.

If you had two tractors, they could plow in formation using GPS, but both with not be precise to the actual ground.

 

I recall somebody made a semi-autonomous flight system for a fixed-wing rc aircraft before the AHRS systems got small and cheap and did wing levelling by having a separate GPS receiver in each wingtip.   It seemed to work pretty well.

[Steve0715]

i was in Albuquerque a couple of weeks ago and they had this drone show.  The guy with the show was staying at my hotel.  He said the drone work on gps.  Each drone is autonomous and each has its own program.  He has 20 drone substitutes.  To insert a substitute he simple tells that drone which one it is.  Amazing to watch.

[Pinecone]

18 hours ago, A64Pilot said:

No, if you took a Gps and put it on the ground in a fixed spot and had it record positions, it would record a circle of dots, the circle would represent of course positional data, as different GPS’s don’t take positional data at the same precise instant, they will record different locations, if you could precisely synchronize them , then they would agree I guess. The tractors would weave about and not together. The inaccuracy isn’t constant, it’s random, with SA back in the day it added 50 meters of error horizontally and 100 meters vertically, without WAAS I think best accuracy possible was 10 meters. Neither would steer a tractor, that needs to be within inches, so enter Differential GPS which takes you from the average 15 meters to roughly 2 cm, good enough for the tractor to be more accurate than any farmer, this increased crop yields significantly because the GPS tractor being more accurate would plow more rows in the same acreage.

 

Yes, because you are referencing to the ground.

Look at it this way, put two GPSes on the ground a foot apart and record the position reported at the same time on each.  Yes, you will see two circles of dots, but any two dots recorded at the same time will show 1 foot apart.

[N201MKTurbo]

5 hours ago, Pinecone said:

 

Yes, because you are referencing to the ground.

Look at it this way, put two GPSes on the ground a foot apart and record the position reported at the same time on each.  Yes, you will see two circles of dots, but any two dots recorded at the same time will show 1 foot apart.

You are correct, except if you plot GPS positions, you get a fuzzy ball, not a circle. I have done it.

[N201MKTurbo]

GPS has no idea where the ground is. The satellites report their positions (orbits) relative to the center of the earth. 
 

The receivers calculate their positions relative to the received satellites. This position is corrected by the broadcast satellite position models. This gives a position relative to the center of the earth. The intersection between the vector from the center of the earth and the standard geoid (mathematical model of the earths surface) is used to calculate the Lat / Lon values. The MSL altitude is the height of the receiver vector above the geoid.

[Pinecone]

2 hours ago, N201MKTurbo said:

You are correct, except if you plot GPS positions, you get a fuzzy ball, not a circle. I have done it.

True.

In the early days of using GPS for surveying, they would place it and let it run for hours.  And then average the reported location to find the actual location.

And the height above ground was controlled to remove that variable.

 

[EricJ]

17 minutes ago, Pinecone said:

True.

In the early days of using GPS for surveying, they would place it and let it run for hours.  And then average the reported location to find the actual location.

And the height above ground was controlled to remove that variable.

 

That was dithering for Selective Availability, which was turned off in 2000 or so.    Everybody gets non-dithered performance now, although I suspect the gov can always selectively degrade the system whenever they want.

There are still error sources that are essentially random, from things like atmospheric effects, multipath, and implementation issues.   Atmospheric effects will be common to all the units in a swarm, and most swarms outdoors in the open won't have significant multipath errors, which would be independent across units more than half a wavelength apart.   So there are still random, independent error sources, but apparently they're small enough that the swarm displays work well.