Re: [Paparazzi-devel] Port airborne code to Linux

[marcus.wolschon]

On Mon, 09 Mar 2009 10:02:48 +0100, "Achim Walther" <address@hidden>
wrote:
>> Is it time to start moving the paparazzi airborne code to Linux? What do
>> you 
> 
> Yes and No. :-)

Okay, not "move to" but more like "establish as another hardware-platform".
Bad wording. The tiny-boards are a great piece of hardware and for most
people
here seem to be the definite way to go.

> I cannot speak for Paparazzi here because I'm not implementing it - just
> signed up to this list to be kept updated. I've been developing a Linux
> based autopilot for the last 4 years and reached a stable traffic circuit
> last sommer. Like Marcus I'm using a Gumstix onboard computer together
with
> an improved Rotomotion IMU. 

Cool. Do you have a website where I can have a look at your aproach and
learn from it?

> On one hand Linux has the advantage to provide an uniform and convenient
> environment for rapid prototyping. Connecting a GPS is just a "cat
> /dev/ttyS3" and connecting I2C is easy as well. Using SD cards there is
> lots of disk space for extensive logging. I am currently testing the
fusion
> of SRTM data with altitude information - SRTM files for Germany take
about
> 200 MB. Bluetooth or Wi-Fi provide standardized communication (think of
GPS
> coupled directional antennas).

As I am using OLSR-mesh-networking I for example get relaying of downlink-
messages from one drone to another (others on weels or stationary) to
nodes of a local Freifunk-Community over the internet to the GCS for free.
Think drones or drones and Freifunk-activists acting together. Now that
sounds like a fun field of research to play in. :)

> On the other hand Linux is a bit oversized just to be used for an
> autopilot. Even with loads of IMU calculations my 166 MHz Gumstix
(reduced
> clock rate due to GPS interference)

Can you tell me more about your setup and the GPS-interferences?
Where did you place the GPS-antenna and what antenna did you use?
Are you using the GPSstix?

> has a processor load not greater than 20 percent.

I don't see much added weight and power-consumption so I don't actually
care
about not using the full potential of the platform
as long as it also gives me the highly advanced communication- and sensors-
possibilities the board has compared to a bare-bones microcontroller.

> Adding payload software to the system might result in timing
> issues so the out-of-the-box OS might need to be tuned into an RTOS.

The timimg-issue is just a software-issue and would not merit adding
the weight, size, cost and complexity of a complete second computer
with power-supply, gps-access and access to the downlink/uplink and
communication with the autopilot.
Getting the paparazzi feedback-loop initiated by a realtime-event
should not be hard to code while still keeping the codebase independent
of the board used (tiny, old atmel-boards, simulation, linux-userspace
and rtlinux if needed).

Of cause I'm using a very stable and slow airframe (trying to get
fully autonomous landing using ultrasound working some day) so I don't
concern myself with anything below a tenth of a seconds like a helicopter
would need to.

Marcus