[OpenRelief Developer] Raspberry Pi to APM2 Communication - Producer-Server-Consumer Model

[Shane Coughlan]

Hi George
On Jun 13, 2012, at 11:31 AM, George Harkin wrote:
> Just joined up to the group today, but I've been working on a similar platform to yours for the last few months on my own.
> Currently I've got an APM2 set up with an FPVRaptor.  It has been a great process getting everything up and running with it, and I'm now confident that it is stable enough to start development with.  It is also set up with FPV, but that might just be for fun ;)
> I've also just received my Raspberry PI, which I felt would be a great small form factor PC to utilize for both onboard image processing and as a publisher to communicate current telemetry and status up to a central server.  
> To that end, I've acquired a U600 USB GSM modem for data link, and will be setting up the UART port on the RaspberryPi to communicate the Mavlink data stream to a server central server.  I hope to then either adapt the ArduPilot Mission Planner to read data streams from that server and issue mission commands back to the Drone.
> If you'd like to help work on this project above, send me a note.  If you think the above project is in the wrong direction, let me know in this thread and we can redirect my effort where it is needed.

I don't think you platform is wrong. At this stage, we are all exploring. 
There are a few small differences between where OpenRelief is currently going and where you are going, perhaps related to the choice of airframe and to your use of GSM, but we are talking marginal stuff here. I suggest we work together as much as possible. In practice everything we do together can be used on both our systems, and we both can test to half the development time. Indeed, as far as I can tell what we want to do on communications is pretty much identical. We want the drones to have a pilot, a computer and get images, and we want to be able to send them on missions and - conditions permitting - monitor them in real time via networks.
In case you are wondering, our current prototype airframe and our future open hardware airframe are intended to be stable, slow and carry a reasonable amount of weight for their size.
> A side project, I'm also working with python-opencv to get a downward facing infrared camera running with some human recognition algorithms.  I think this would be _very_ useful in both search and rescue operations, or simple affected persons counts.  
> I've specked out small form factor infrared cameras, and the lowest I could find was in the $2000 to $3000 range ( see http://www.drs.com/products/RSTA/Tamarisk.aspx ).  If you know of a lower cost small form factor infrared camera, please let me know.  I might be able to talk to drs and see if they would be willing to chip in a development camera in prospect of future sales.

Hopping forward a little, infrared cameras are good, but they are expensive. After we get the production basic design completed, we can start to put energy into "expansion modules" like infrared support. Thinking down the line, there should be a range of technology that you can deploy on the ground or in the air that knits together neatly. But first things first, we need to solve the basic airframe, basic avionics, basic vision and basic drone management challenge. :)
Shane