FPV control via 3g/4g cellular connection

[David Janssens]

We recently developed a controller to pilot drones/cars/boarts/etc via 3G cellular connection, in order to achieve "unlimited" control range.
The code and schematics are all open source and available here: https://github.com/nfco/cellpilot

We are currently testing it using hand-soldered PCBs and 3d-printed cases.
We also just started a campaign to fund the first production run:
https://www.indiegogo.com/campaigns/cellpilot

Any feedback would be appreciated to guide the development of the project.

 

[David Janssens]

I apologize the IGG link was broken, the proper link is https://www.indiegogo.com/projects/cellpilot/x/12417937#/

Here is a video of a 64kbps flight test this morning:

Anyone else here trying to do FPV via cellular?

 

[ProfEngr]

I was toying (ha ha) with the idea (work-related) for when I become more proficient and step up to a fuel-powered quad, but my current level of flight experience is novice. I have a prototype bi-rotor I built during my college days that I never put the finishing details on, but cellular tech back then was a bag phone. My biggest concerns with this tech now are FAA/FCC not playing nice and craft location/recovery in the event of a mechanical failure. From what I've seen on the web, all the pieces are available if someone wanted to put them together to make a system such as this operational. Network interface lag could be another issue via 3G/4G, but I haven't really made the time to test it.

 

[David Janssens]

Thanks for your reply!
Regarding the network lag, I first connected to the drone using STUN (3G devices are usually behind NAT) in order to minimize delay.
After that I tried making a simple UDP relay that runs on a server (the relay is 20 lines of python and on github), to improve reliability in case the 3G IP address changes.
After some tests, I think network lag is not really a problem, it's not to difficult to get it around 100ms and low enough to fly.
I would also be interested if anyone has ideas in order to improve flight safety. Currently the drone enters failsafe mode and returns home in case the 3g connection is broken or if the raspberry pi crashes, but I'm sure there are still some improvements that could be made.

 

[ProfEngr]

I've often wondered how well the 'return home' feature works if you've managed to fly around/behind a higher ground elevation. Does it just fly a vector back to base? That scenario could result in an unintended crash.

 

[David Janssens]

AFAIK, the return home feature first elevates the drone to 20-30m before coming back. So if something higher is in the way this is a problem. Not sure how to avoid this, maybe the simplest solution is to increase the evelation to 100m or something like that. Or it may be possible to have feature where the drone follows the same path to go back that it used to leave.