1
0
mirror of https://bitbucket.org/librepilot/librepilot.git synced 2024-12-10 18:24:11 +01:00
Commit Graph

75 Commits

Author SHA1 Message Date
James Cotton
9c512261e6 When in indoor mode reset the baro offset to zero 2012-07-22 01:02:14 -05:00
James Cotton
5649813c2b Make sure to always compare BiasCorrectedRaw against its enum. Also reversed
the order of TRUE,FALSE in the definition to make it safer anyway.
2012-07-21 17:24:58 -05:00
James Cotton
d36663dbbd Treat the GyroBias UAVO like a state estimate of the actual gyro bias so now we
subtract that from the raw sensor readings to get the Gyros UAVO value
2012-07-21 15:14:45 -05:00
James Cotton
3e33bb49a2 EKF gyro bias into the object now working propely 2012-07-21 15:11:04 -05:00
James Cotton
02dfa7bd82 Change how the updated settings in attitude are changed to make it easier to
initialize them all
2012-07-21 14:31:14 -05:00
James Cotton
3f4706ad4c Make it a critical error when an invalid attitude algorithm is selected 2012-07-21 14:00:52 -05:00
James Cotton
69057a1373 No reason to get homeLocation every cycle 2012-07-21 13:57:25 -05:00
James Cotton
7cdf47c1d4 Make the sensors code only apply the GyroBias from the UAVO which is now coming
consistently only from the attitude module
2012-07-21 13:51:20 -05:00
James Cotton
e5bd999975 Process the settings updates more discretely in revo attitude 2012-07-21 13:50:16 -05:00
James Cotton
797a4def6a Do not use the data from the magnetometer if it contains NAN 2012-07-21 11:43:31 -05:00
Corvus Corax
be72d24c5f Attitude: Fixed calculation for baro offset adjustment time 2012-06-03 09:52:24 -05:00
Corvus Corax
b06b51f1b2 Revolution/Attitude: Added offset calculation for barometric altitude 2012-06-03 09:52:14 -05:00
James Cotton
836e34a483 Change revo sensors to update at 500 Hz and make sure in all cases it waits for
the relevant queue to provide data.  Also disable the quaternion stabilization
flag.
2012-05-12 13:12:56 -05:00
James Cotton
2ca3cde83f Improve EKF initialization. For some reason we must read an extra mag sample
when power up value of attitude algorithm is EKF.  I don't understand why the
first mag value is bad.
2012-05-12 13:00:51 -05:00
James Cotton
98655aacc8 Fix initialization of the complimentary filter where it didn't get the right
initial dT
2012-05-12 13:00:50 -05:00
James Cotton
6c0dc80978 Shuffle more stack sizes around 2012-05-05 11:33:47 -05:00
James Cotton
fb53d5621a Change the magKp term (hardcoded still) for revo complimentary filter so it
tracks magnetic north when homelocation is set.
2012-04-22 19:01:41 -05:00
James Cotton
2890fcdc9f When in complimentary filter mode use GPSVelocity so event system doesn't have
warning.
2012-04-22 19:01:18 -05:00
James Cotton
a7ba6d96c1 Allow guidance to run purely off raw GPS data 2012-04-15 16:23:34 -05:00
James Cotton
1559eeb2e1 Use GPSVelocity updates when they are available instead of timing based on
GPSPosition
2012-04-15 11:50:56 -05:00
James Cotton
1a98a46606 Merge branch 'revolution_ubx' into sim 2012-04-14 12:10:23 -05:00
James Cotton
5d160860a3 Fix from Corvus for gyro bias from EKF 2012-04-14 12:09:12 -05:00
James Cotton
136bbe3814 Update the EKF code to use the GPSVelocity 2012-04-09 19:03:21 -05:00
James Cotton
414e62f14e Revert "Disable baro updates into INSGPS for now. Seems to cause issues."
This reverts commit 04591309770679e7b01c53a6cb54cac1c043f8b6.
2012-04-07 00:58:36 -05:00
James Cotton
fcb9193d25 Disable baro updates into INSGPS for now. Seems to cause issues. 2012-04-07 00:58:36 -05:00
James Cotton
81fcfd45c4 Some work on the code to initialize the INSGPS and allow setting the variance
of the baro
2012-04-07 00:37:42 -05:00
James Cotton
345b6578b1 Fix a possible race condition for complimentary filter on L3GD20 attitude
estimation where the accel queue wouldn't immediately have data available.
Added 1 ms timeout on queue.
2012-04-06 13:59:12 -05:00
James Cotton
ea13536da0 Take quidance settings from simulation and make the INSGPS pick up the GPS
noise from a setting.
2012-04-06 12:59:49 -05:00
James Cotton
6df9691609 Bring back the original module files and remove comments from FreeRTOS
debugging
2012-04-03 09:16:30 -05:00
James Cotton
fbd8695d7d Check in WIP for osx simulation 2012-04-03 09:16:30 -05:00
James Cotton
e97c1bc016 Include the module include file before any uavo ones 2012-04-03 09:16:27 -05:00
James Cotton
5b3cc4876e In simulations a higher GPS noise is required for the INS 2012-04-03 03:42:35 -05:00
James Cotton
7f226867c6 Some bug fixes for revo attitude estimation. 2012-04-03 03:42:33 -05:00
James Cotton
af9f36d7df Improve the attitude initialization for all of the modes 2012-04-03 03:42:32 -05:00
James Cotton
3888d6ff69 Changed settings that makes updates smoother 2012-04-03 03:42:31 -05:00
James Cotton
319b53a342 Make sure full initialization routine for INSGPS occurs each time it is
restarted.
2012-04-03 03:42:31 -05:00
James Cotton
372f959c37 Refine some stack sizes 2012-04-03 03:42:30 -05:00
James Cotton
54880c83a0 Fix some task sizes and the memory limits 2012-04-03 03:42:30 -05:00
James Cotton
a83fb019f6 When in complimentary filter mode pass the GPS information straight
through to Position Actual
2012-04-03 03:42:29 -05:00
James Cotton
98b1a21921 Fixup: Shouldn't update gps until home location is set 2012-04-03 03:42:29 -05:00
James Cotton
7a3ec3e173 Some code cleanup to get rid of some warning messages 2012-04-03 03:42:29 -05:00
James Cotton
23625904c5 Remove the now unused ECEF and Rne fields from HomeLocation. Also reenable
calculation of the WMM and HomeLocation from the GPS code for revolution.
2012-04-03 03:42:28 -05:00
James Cotton
71d0180d45 Use the sensor variances from object 2012-04-03 03:42:28 -05:00
James Cotton
ce2d1f94f6 Better hardcoded constants for INS variances 2012-04-03 03:42:27 -05:00
James Cotton
de7fc81795 Fixes to the GPS LLA to NED translation 2012-04-03 03:42:27 -05:00
James Cotton
e32ab9342c Shuffle some stack sizes around 2012-04-03 03:42:26 -05:00
James Cotton
4cadb7ec0e Forgot to initialize revo settings object 2012-04-03 03:42:26 -05:00
James Cotton
0eedaa1250 Change how we convert LLA to NED. Now it is done with a taylor expansion
around the home LLA coordinate to avoid the conversion into ECEF coordinates.
This has the benefit of not requiring double precision math and uses less
operations.

Now we should remove the Rne and ECEF fields from HomeLocation as they are
unused
2012-04-03 03:42:25 -05:00
James Cotton
e8cc7748af Clean up some unused variables in attitude code 2012-04-03 03:42:24 -05:00
James Cotton
a9aa6b696b Add object for setting the fusion algorithm 2012-04-03 03:42:24 -05:00