How to control multiple drones at once?
- Thread starter Roshi
- Start date
Roshi
Member
I may be able to get funding but why does it cost loads? What hardware would be costly here?
I am really interested in this project? If possible please tell me how I can accomplish this? If there are any commercial systems out there that you can recommend please do. Id like to know how these systems work.
I simply imagine a device with multiple transmitters on it, 1 for each drone and a system like a laptop to program each transmitter
I am really interested in this project? If possible please tell me how I can accomplish this? If there are any commercial systems out there that you can recommend please do. Id like to know how these systems work.
I simply imagine a device with multiple transmitters on it, 1 for each drone and a system like a laptop to program each transmitter
Chuck
Well-Known Member
The closest thing to that which I can think of (and the equipment to accomplish it is not normally commercially available), is what was originally called: "MIRV" or "Multiple Indepent Re-entry Vehicle". A form of ICBM (Intercontinental Ballistic Missile) developed by both the United States, and the former Soviet Union. Where one missile carried multiple warheads, each capable of striking a seperate target after launch. For obvious reasons, neither "inventor" chose to make this technology commercially available through your traditional retail electronics outlets....
How many are we talking about?
You could do tagged traffic, or a kinda ADhoc system with a client server relationship.
Either way the use will be limited and the cost and complexity will be higher.
Chuck
Well-Known Member
You also didn't mention if you are trying to control them independently like multiple pilots controlling each individually, or if you're trying to fly them in a group formation. Where they all take off together, head off in the same direction, at the same altitude and speed, turning in formation, and return and land in formation.
Last edited:
You'd need a way to determine relative location to each other to avoid bumping into themselves. NFC could be used I suppose to give warning and make them keep away.
mozquito1
Well-Known Member
I take it is beyond anyone here without real digging.
I just googled a bit & came across a piece of software called Pixiepath.
Sorry I don't know how to link.
Can build & fly kwads but still can't link.
It is a Quadcopter forum but that's a bit more computer programming.
Most of us could build you the drone you need, but I think for tight aerobatics you will struggle with this. From what I seen of multiple drone shows. It is a closely guarded secret. The man hours that goes in & the cost, it's not likely to be open source. You would probably have just as much luck on a programming forum.
They might have an idea but then think I know nothing about drones without digging into it. Sorry it's not something I would be interested in doing myself.
So this is where I wish you all the best in your endeavour. I'm stumped
I just googled a bit & came across a piece of software called Pixiepath.
Sorry I don't know how to link.
Can build & fly kwads but still can't link.
It is a Quadcopter forum but that's a bit more computer programming.
Most of us could build you the drone you need, but I think for tight aerobatics you will struggle with this. From what I seen of multiple drone shows. It is a closely guarded secret. The man hours that goes in & the cost, it's not likely to be open source. You would probably have just as much luck on a programming forum.
They might have an idea but then think I know nothing about drones without digging into it. Sorry it's not something I would be interested in doing myself.
So this is where I wish you all the best in your endeavour. I'm stumped
Chuck
Well-Known Member
Another question is, what do you mean by: "a fleet or swarm"? If you're referring to three-to-five, wouldn't it be easier to just get a group of drone-flying friends together to practice in a military-style flight formation? There is considerable documentation online about those. Such as the US Army Air Forces B-17 bomber formation known as the "Combat Box". Or the fighter swarms used by Japanese Naval fighter pilots in the Pacific. But if you're after something involving ten or more drones simultaneously, my next question is: How did you afford to build such an armada in the first place?I take it is beyond anyone here without real digging.
I just googled a bit & came across a piece of software called Pixiepath.
Sorry I don't know how to link.
Can build & fly kwads but still can't link.
It is a Quadcopter forum but that's a bit more computer programming.
Most of us could build you the drone you need, but I think for tight aerobatics you will struggle with this. From what I seen of multiple drone shows. It is a closely guarded secret. The man hours that goes in & the cost, it's not likely to be open source. You would probably have just as much luck on a programming forum.
They might have an idea but then think I know nothing about drones without digging into it. Sorry it's not something I would be interested in doing myself.
So this is where I wish you all the best in your endeavour. I'm stumped
Jeff Neese
Well-Known Member
Chuck
Well-Known Member
Intel can easily afford a lot more than the average hobbyist drone pilot can. Including 500-drone air wings.
Jeff Neese
Well-Known Member
The OP said he may be able to get funding.
We obviously need more details, but if it requires either one pilot controlling several or one auto pilot controlling several then the only real option would be to have a centralized computer tell multiple multirotors what to do based off it's centralizes auto pilot or based off from user input commands. You could use it to fly formations or to do something like an automated flight system where multiple aircraft are going on completely different flight plans. You could either try to somehow do them on different frequencies (does not scale well), or to use a protocol that allows you to tag traffic to whom it concerns type deal like an older hub network where all PCs were on the same bus.
Chuck
Well-Known Member
We also need to know what purpose for a drone formation of dozens or hundreds of drones, to be controlled by one operator.
Roshi
Member
I'm looking to control about 5-10 drones. I agree with the centralized system. Auto pilot sounds best as one pilot could not control 10 drones at once. Although a flexible program that allows flight plan changes and manual over-ride options for each individual craft would be preferred. I'm interested in controlling fleets for the many applications it has to offer. Multiple Drones Filming the same scene from different angles, Surveying areas of crime incidences, Formations for 3d mapping, entertainment/recreation..
I'm also very interested in the idea of data traffic being directed to which craft it concerns instead of having multiple txrxs on different frequencies. Maybe the drone can receive data from the centralized station, download and save its mission/commands and then execute it without unnecessary communications from Central Station.
I would still like to know why this system would be so costly as i'm interested in DIYing. If I am confident that this is doabe I can probably invest 10k-15k on this project. Some kind of GCS with strong UAV swarming capabilities is something i would be very interested in
Gyro Doctor
Electronics Tech for over 45 years
... The concept isn't that complicated really, but the application (as far as I know) would require custom programming and is extremely complicated.
a.) Controlling the swarm: A single radio controller interfaced to a laptop that either drives multiple TXs or a single TX enabled with high speed frequency hopping.
The software for this would likely need to be custom programming.
b.) Distancing the formation between themselves: An independent transponder per drone arrangement with RSSI that would be tunable from the TX (relative to signal
strengths the formation could "close ranks" or spread out further from each other), which is again, a custom application. Some higher priced drones come with collision
avoidance systems on board (ultrasonic and infrared) that could possibly be adapted for tunability to achieve the same results. Envision simply turning a knob on the radio
controller to adjust the distance each would keep from the other.
c.) Collective vs individual control: Note that either the RSSI or collision avoidance system scenarios are simply distance related, without regard to actual position.
Since the swarm flies in 3D space a shared precision barometric altitude feedback loop would also be necessary to keep them all in vertical alignment too.
Furthermore the altitude setting would need to be dynamic, rather than static (as in currently available "off the shelf" altitude hold systems) in order to keep each
individual able to compensate for deviations from the intentions of the control inputs. In other words, each quad might respond to said input (up or down) but wind up
actually moving either not enough or too much, so all altitude change commands would need to be referenced to a commonality (actual altitude) in order for the swarm
to stay in formation while responding to such inputs.
In summary, each quad has to be not only aware of it's location relative to the others, but also able to react to control inputs with compensation algorithms incorporated
to keep it's self a part of the collective swarm and not stray from the formation despite inevitable variations from it's responding to control inputs.
GPS awareness could possibly be substituted for RSSI and barometric referencing for general formation keeping, but this would be less precise for close rank flying. Ideally
a combination of both would enable both a wider formation capability (GPS) as well as precision grouping (RSSI/Baro).
Another aspect of consideration is the approach to be taken regarding the swarm's control mechanisms, which is to ask: Shall each drone be an identical autonomous entity,
or might a "Client/Server" configuration be warranted ?
There are pros and cons to each approach:
The client/server configuration would likely be a (somewhat) simpler and much less expensive (single transmitter for server with "blind" clients) way to create a swarm, with
the "server" being a lead quad that all the others simply follow about while trying to keep their relative distance from while matching the same altitude. But the "clients" would
then sacrifice the ability to perform the same maneuvers of the server and without some additional complex programming on board (enter vector algebra 101) only be able to
make basic azimuth and altitude corrections (forget about synchronized banked turns and such). Also, anything going wrong with the server would lead to a "daisy chain" effect
that would wipe out the entire swarm.
Individual autonomous quads acting as a collective would be more complicated and expensive, but really much more impressive, since they would all perform the exact same synchronized maneuvers (banks, flips, anything) while their onboard programming maintained relative distance and altitude as before, with the additional bonus of any single
quad failing (even falling out of the sky) being inconsequential as the others would simply sense the gap in formation and adjust accordingly.
a.) Controlling the swarm: A single radio controller interfaced to a laptop that either drives multiple TXs or a single TX enabled with high speed frequency hopping.
The software for this would likely need to be custom programming.
b.) Distancing the formation between themselves: An independent transponder per drone arrangement with RSSI that would be tunable from the TX (relative to signal
strengths the formation could "close ranks" or spread out further from each other), which is again, a custom application. Some higher priced drones come with collision
avoidance systems on board (ultrasonic and infrared) that could possibly be adapted for tunability to achieve the same results. Envision simply turning a knob on the radio
controller to adjust the distance each would keep from the other.
c.) Collective vs individual control: Note that either the RSSI or collision avoidance system scenarios are simply distance related, without regard to actual position.
Since the swarm flies in 3D space a shared precision barometric altitude feedback loop would also be necessary to keep them all in vertical alignment too.
Furthermore the altitude setting would need to be dynamic, rather than static (as in currently available "off the shelf" altitude hold systems) in order to keep each
individual able to compensate for deviations from the intentions of the control inputs. In other words, each quad might respond to said input (up or down) but wind up
actually moving either not enough or too much, so all altitude change commands would need to be referenced to a commonality (actual altitude) in order for the swarm
to stay in formation while responding to such inputs.
In summary, each quad has to be not only aware of it's location relative to the others, but also able to react to control inputs with compensation algorithms incorporated
to keep it's self a part of the collective swarm and not stray from the formation despite inevitable variations from it's responding to control inputs.
GPS awareness could possibly be substituted for RSSI and barometric referencing for general formation keeping, but this would be less precise for close rank flying. Ideally
a combination of both would enable both a wider formation capability (GPS) as well as precision grouping (RSSI/Baro).
Another aspect of consideration is the approach to be taken regarding the swarm's control mechanisms, which is to ask: Shall each drone be an identical autonomous entity,
or might a "Client/Server" configuration be warranted ?
There are pros and cons to each approach:
The client/server configuration would likely be a (somewhat) simpler and much less expensive (single transmitter for server with "blind" clients) way to create a swarm, with
the "server" being a lead quad that all the others simply follow about while trying to keep their relative distance from while matching the same altitude. But the "clients" would
then sacrifice the ability to perform the same maneuvers of the server and without some additional complex programming on board (enter vector algebra 101) only be able to
make basic azimuth and altitude corrections (forget about synchronized banked turns and such). Also, anything going wrong with the server would lead to a "daisy chain" effect
that would wipe out the entire swarm.
Individual autonomous quads acting as a collective would be more complicated and expensive, but really much more impressive, since they would all perform the exact same synchronized maneuvers (banks, flips, anything) while their onboard programming maintained relative distance and altitude as before, with the additional bonus of any single
quad failing (even falling out of the sky) being inconsequential as the others would simply sense the gap in formation and adjust accordingly.
You forget about another option in the client/server configuration.
You have all of the quadcopters report to the ground station (central controller, could be a laptop) with whatever sensor data they have such as GPS, and whatever assortment of sensors you decide for improved flight or mission payload, and they add an identifier in the header to tell which quadcopter it came from. The ground station could then process that and position each quadcopter on the model it has of where they are. It would then just tell each individual quadcopter what, where, and how and they follow it blindly. The advantage is that you are less likely to have impossible situations when troubleshooting to figure out simple problems as all problems generally will come down to the ground station. Also it would be a lot cheaper and you could model them to fly however you decide and not worry about them reacting to unforeseen problems. As for collision avoidance you would also benefit of the sensors giving data to one central location and all of them being able to react from that.
You could even do this over existing protocols such as the all powerful 802.11 (Wifi). Would just need a good transmitter and antennas to get a reasonable range.
You have all of the quadcopters report to the ground station (central controller, could be a laptop) with whatever sensor data they have such as GPS, and whatever assortment of sensors you decide for improved flight or mission payload, and they add an identifier in the header to tell which quadcopter it came from. The ground station could then process that and position each quadcopter on the model it has of where they are. It would then just tell each individual quadcopter what, where, and how and they follow it blindly. The advantage is that you are less likely to have impossible situations when troubleshooting to figure out simple problems as all problems generally will come down to the ground station. Also it would be a lot cheaper and you could model them to fly however you decide and not worry about them reacting to unforeseen problems. As for collision avoidance you would also benefit of the sensors giving data to one central location and all of them being able to react from that.
You could even do this over existing protocols such as the all powerful 802.11 (Wifi). Would just need a good transmitter and antennas to get a reasonable range.