Hey, not even sure where to ask about this so apologies if this is the wrong spot.
I’m wanting to get multiple drones to follow a predetermined path, preferably by passing accelerometer data from a lead drone to a follower drone, and then from follower to another follower, etc. I’ve seen something vaguely similar done with Parrot drones, but some searching lead me to believe that those drones may no longer be available / no longer supported.
Can anyone help point me in the right direction on how to do this? I can help clarify if there’s more information needed to answer my question, I’m sure there’s something I’ve either overlooked or not considered.
Ah! Yes that’s very close to what I’m looking for.
The only distinction is that it looks like these implementations are for GPS-based formation flying, while what I’m wanting to do is polling data (positioning data for each drone, primarily accelerometer, secondarily GPS) and using that to control the drones from a central computer with my own code in real time. The plan is to have one manually controlled drone using a radio transmitter, and have all remaining drones follow the path of the drone preceding it as best it can.
Is there any API documentation on how to do this? I’m assuming the error / drift rate will be fairly high, but I’m hoping it will be usable for 30-60 second demonstrations before needing to be reset.
I’m planning on starting with 3 total quadcopters, and then if the code appears to be working, scaling it up according to how much I can afford and how bad the error rate grows per additional drone. (5-7 would be an ideal number, IMO)
Bumping to see if any ArduPilot devs might know where to start. I assume I’ll have to build some of this functionality myself, but I don’t want to accidentally re-develop existing functionality just because I didn’t know it existed.
The swarming stuff linked above is close, but I’d want to hook into whatever those 3rd party applications are hooking into myself.
AFAIK there is no such functionality, and giro/accel navigation is not feasible with the current MEMS sensors, since the drift is quite large and position error grows exponentially once there is no GPS.
30-60 sec is optimistic at best.
As per my knowledge inertial only navigation with copters is not possible with the current sensor sets.
That’s unfortunate. It sounds like I could still accomplish this project by using GPS and triangulation techniques to error correct, but not without spending significant effort.
I’ll have to put this one on ice for now I guess, and hope that either more accurate accelerometers are developed, or I get familiar enough with control systems to build my own.
