Re: [Paparazzi-devel] Using Thermopiles for AHRS on Lisa/M instead of IMU

[Christophe De Wagter]

You can simply use

<ahrs type=infrared>

<imu type=aspirin>

-> No other ahrs like DCM, however the imu.calibrated_float.p might still need to be copied to estimator_p

And just use the roll gyro from the aspirin (then the aspirin is a bit of an overkill). For the pitch damping it is more difficult as you need a combination of the pitch and yaw gyro that depends on the roll angle and this was never released in paparazzi.

___

The other option of integrating gyro's (e.g. DCM filter) and using the thermopiles + GPS as feedback is also possible. It is as simple as replacing the accelerometer+GPS vector with a thermopile attitude vector in ahrs_dcm_correct. It would mainly improve the slow response of thermopiles, but will NOT improve errors in thermopiles. On the contrary, if your thermopiles have a roll offset and you make circles the filter will constantly estimate an varying gyro bias which might even make the biggest attitude error even slighly larger than with thermopiles only. 

On the other hand we have had nice results with very very slow accelerometer + GPS estimates to correct thermopile biases or even scale. That seemed to work as a charm. 

Finally, we have had unstable IMU attitude estimates quite a few times the last years (turbulent weather, broken propellor tips, too high control loop gains, too aggressive manual flight, ... ) and then the plane would simply crash if you left it in auto2. Having thermopiles here would be a nice complement (if we would find a good parameter to know which attitude source should be used)

Conclusion:

Thermopiles + Gyro = performance increase

Thermopiles + Accelerometer&GPS = bias removal

Thermopiles + IMU-AHRS = extra error recovery options

Certainly, many more combinations are possible. 

-Christophe 

On Thu, Apr 5, 2012 at 10:35 PM, Chris Wozny <address@hidden> wrote:

Christophe,

Something you said made me curious. Is it possible to use the Aspirin
IMU solely for gyro stabilization if thermopiles are being used for
attitude estimation? I know we've used 3-axis gyros in the past with
Tiny v0.99 for roll/pitch/yaw dampening while using IR, but I don't
know if this is any different since the gyros are on the Aspirin IMU.

Best,
Chris

On Tue, Apr 3, 2012 at 12:05 AM, Christophe De Wagter
<address@hidden> wrote:
> Hi,
>
> The default assignment of adc should be written in
> conf/board/lisa-m.makefile
>
> There might still be the issue that ADC_1 is called 0
> USE_0 instead of USE_ADC_1
>
> For the gyros: the value of the gyro is only distorted a few percent. So for
> simple stabilization like a roll damper the aspirin is good 3noug even with
> vibrations. The problems start when you start integrating, which is not
> needed for damping but required for attitude. Then all errors accumulate.
> And a gyro error of 1% (sounds like a small error) on a 2000 degree/sec gyro
> gives 20 degree of additional attitude error PER SECOND. This is corrected
> with accelerometers. However the accelerometers also measure vibrations.
> These can be several times bigger than gravity. This means that at times the
> accelerometer even could measure that the earth is up. And low pass
> filtering them too much gives latency. And that again gives correction
> errors:  e.g. if you suddenly roll 90 degrees, but the accelerometer follows
> with a delay, initially the filtered accelerometer will measure a huge error
> and estimate this is due to the gyro bias.
>
> Did you try to mount the lisa on 4 corners of 3x3x10mm pieces of tempur foam
> and use very thin flexible cables?
>
> On Apr 3, 2012 5:06 AM, "Chris Wozny" <address@hidden> wrote:
>>
>> All,
>>
>> For reasons outside of my power (trying to dampen vibrations in tiny,
>> tiny aircraft leads to massive hair loss and frustration) I've got to
>> resort to using IR for the AHRS on our aircraft instead of Aspirin
>> v1.0. We're using Lisa/M and I was wondering: are there any default
>> assignments for which ports to use for thermopiles like with Tiny? The
>> wiki says there should be, but I have a feeling Lisa was designed
>> without thermopiles in mind since an IMU would work much better.
>>
>> Also, why do analog gyros (more specifically Futaba Mini Piezo gyros)
>> work great for stabilizing the same plane with the same vibrations
>> that cripple the use of Aspirin v1.0 for AHRS?
>>
>> Best,
>> Chris
>> _______________________________________________
>> Paparazzi-devel mailing list
>> address@hidden
>> https://lists.nongnu.org/mailman/listinfo/paparazzi-devel

_______________________________________________
Paparazzi-devel mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/paparazzi-devel