The Paparazzi project is distinctive for using and IR sensor like the FMA Co-Pilot, rather than a gyro-based IMU, for stabalization. It's computationally a lot easier and I'm told works very well. That's the open source autopilot that is used by many of the contestants in the European UAV contests.
If anyone wants to start a Paparazzi project and document it here, I'd be happy to feature it on the front page. It's at the high end of our range, but not beyond it.
I'm working with paparazzi.
The airborne was quite a step to learn but now it's working.
(I had to add not only a new airframe but a new build-target with different sensors.)
You can get it running on their arm7-based system, on atmega-microcontrollers and now even as a linux-usermode-program.
It works with 2 or one controllers (separate servo-control and navigation in 2 controllers or 2 processes in one controller).
As you can see, the build-system is quite flexible and hardware-independent.
There are also multiple modems to choose from for telemetry.
The ground-station looks very cool and in good for controlling multiple airframes but the problem it, that it is written in ocaml.
(The airborne code is plain C.) Thus the ground-station is hard to modify unless you spend a lot of time to learn the language first.
You are no longer right about the gyros as there is code to support these and some people are controlling
quadcopters instead of fixed-wing-aircrafts with it. The usual drone will still have IR-thermopiles.
It also featues a booz-based simulation-environment, to test yout autopilot in flightgear.
Navigation requires a DATALINK to upload the flight-path from the ground-station
to the airframe.
However you can use a serial cable on the ground to do that and then fly without
a data-link. Once the flight-plan is in the plane it will execute it.
The flight-plan can not only contain way-points but is written in a touring-complete
xml-language. So you can define what the airframe can do e.g. uppon low-power
or if a waypoint cannot be reached in time, if there is a strong wind comming up
or if an ultrasonic sensor gives a specific reading, indicating the real altotude above
ground while landing autonomously.
It is only possible to change the position of waypoints mid flight.. I have been working on paparazzi for over a year now and actually was part of a UAV team which took second at the student UAS in MD this summer. Here is our website if anyone is interested http://www.engr.usu.edu/wiki/index.php/OSAM .. Overall paparazzi is a great system which i really like using
Also the team i am working with has added support for the use of an IMU and the interface for a gumstix module.
Also the team i am working with has added support for the use of an IMU and the interface for a gumstix module
Can you tell me more about the interface for Gumstix?
I am part of a University UAS team and we have been left gumstix verdex motherboard with memory and console from last years team. Im trying to find out if we can use this with paparrazi or need to purchase new hardware!
You can find the (german) project-documentation on http://ccc-fr.de/index.php/Dronenbau .
I have a gumstix, 4-IR -board, 2-IR-board and a working airframe.
However I did not yet manage to solder the very tiny pitch of the I2C ADC I wanted to use.
On the Gumstix-Mailing-List however there is a new post from today on using the 2 ADC-channels
on the Gumstix-board, so I can do without the external ADC.
* a virtual serial-port connecting to GPSD is included
* pararazzi can do NMEA now (to use the GPSStix)
* downlink via UDP is included
* TCP/IP via mesh-networking (OLSR) works
* Backup-Link via Bluetooth works
* write the required code for the ADC (trivial now)
* test the servo-code (using gpio-pins on the gumstix and the internal timers)
* find out the format of the messages to write the uplink-via-UDP -code.
* find a servo-controlled switch to switch the engine+steering -channels from the AP to RC-receiver for manual override
* I need a stable power-supply for the gumstix.
How did you do the interface with the gumstix-module?
As far as I know booz is a method of looping Paparazzi through flightgear as a simulation-environment.
I was and am quite busy with 2 other free software projects (the Traveling Salesman navigation-system
for OpenStreetMap and jGnucashLib) so my drone is still only flying RC because I did not
come around to solder the ADC for the IR-sensors.
The controls work very much like a plane, so in simulation there was no change of any basic rule
required. Just tuning. I think that will change once I get down to automatic landing and of cause
it reacts very slowly to changes compared to a plane.
On the Gumstix-Mailing-List it was announced that there will be a gumstix-board for UAVs
published tomorrow. So maybe I can use that one to get around my little adc-soldering-issues