rooty/LibrePilot
1
0
Fork 0
mirror of https://bitbucket.org/librepilot/librepilot.git synced 2024-12-01 09:24:10 +01:00
Code Issues Releases Activity
1,671 Commits 5 Branches 8 Tags 266 MiB
c9752a6a3a
Commit Graph

66 Commits

Author SHA1 Message Date
peabody124
0ea95fbad1 AHRS: Separate covariance prediction from state prediction to get updated state out of spi a few milliseconds sooner. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1675 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-17 18:15:00 +00:00
peabody124
92c36114f3 AHRS: No longer reinit after a comms glitch. Allow time for EKF to converge (2-3 seconds) after power up. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1674 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-17 18:14:20 +00:00
peabody124
832b61ab86 OP-117 AHRS: Binary streaming of the raw data within EKF. Doesn't keep up while running this but does give accurate data that went into algorithm. Once we can dynamically change sampling rate will be good. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1655 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-16 05:26:14 +00:00
peabody124
885c966e53 AHRS: Added a tracker for number of ADC updates the EKF misses. Also updated the Telemetry autogenerated files to accomodate the 11500 baud rate. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1652 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-16 02:59:00 +00:00
peabody124
ff1b303083 AHRS: Only do EKF correction with mag when new data available 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1651 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-16 02:58:56 +00:00
peabody124
19e80da928 AHRS: Update UAVObject with information about AHRS rate 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1650 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-16 01:54:21 +00:00
stac
cdb8aa37a0 ahrs: Only read mag data when it has been updated 
The AHRS mainloop was reading the mag data on every
loop regardless of whether new data was actually
available.  Now that the MAG_DRDY signal is monitored,
we can read only at the rate (10Hz) new data is actually
produced by the sensor.

This also enables future improvements that will remove the
filtering work that is still being done on the mag data on
every iteration.

git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1630 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-15 14:21:05 +00:00
stac
55d90970ed hmc5843: hide internal #defines and functions 
Way too much was being exposed in the API for the
HMC5843.  This commit properly hides internal
details of the driver.

git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1629 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-15 14:21:04 +00:00
stac
139da048f6 ahrs: remove unused debug code 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1620 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-15 14:20:51 +00:00
peabody124
c5fe985a45 AHRS: Gets timing working correctly to compute the EKF rate and idle time in real values 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1615 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-14 07:03:36 +00:00
peabody124
0875dc8442 AHRS: First pass refactoring the ADC code into another function. Still want to clean up a few constants make the downsampling go into a smaller circularbuffer nicely. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1614 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-14 07:03:34 +00:00
peabody124
1051d894b9 AHRS: Merged in some new code from Dschin that expliclity writes the prediction terms in the covariance matrix. Also made changes to compile with -Os. Now runs around 125 - 150 Hz depending on corrections used (this will become adjustable from GCS shortly). 
Squashed commit of the following:

commit 897e3365ab6c762584c34238efd12dc144e72af0
Author: James Cotton <peabody124@gmail.com>
Date:   Mon Sep 13 22:15:21 2010 -0500

    AHRS: Getting it running nice and quickly.

commit 53d72fb849164730c89873cab81c9e09be0b112b
Author: James Cotton <peabody124@gmail.com>
Date:   Mon Sep 13 15:57:46 2010 -0500

    AHRS: Make more variables volatile for -Os.

commit 4fa90821b5d3d8c2ea5e67756313a8a63c0eeb8c
Author: James Cotton <peabody124@gmail.com>
Date:   Mon Sep 13 15:11:31 2010 -0500

    AHRS/Makefile: -O0 not added to allow me to try various space flags that don't break code.

commit 5bd17a304dc28dc271c2f3e8c6cb8ad4a830404f
Author: James Cotton <peabody124@gmail.com>
Date:   Mon Sep 13 14:44:05 2010 -0500

    AHRS: Make debuggign USART easy to disable

commit 5453f2a7939492769fe9e9cc822e69c107320670
Author: James Cotton <peabody124@gmail.com>
Date:   Mon Sep 13 14:38:37 2010 -0500

    AHRS: Explicitly expand covariance prediction.  Brings computations from 160 to 90.  Makes it not fit in AHRS though.

commit b6712da7eece2a464a2073d24b77be22bfa47094
Author: James Cotton <peabody124@gmail.com>
Date:   Sun Sep 12 22:59:42 2010 -0500

    AHRS: Make it easy to comment out the I2C/Magnetometers.

commit 8c84d0091f3d6f6bedc7a0224a4ed9cb099b022b
Author: James Cotton <peabody124@gmail.com>
Date:   Thu Sep 9 00:49:50 2010 -0500

    AHRS: Small change to INSGPS analysis tools

git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1613 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-14 06:42:56 +00:00
pip
7da2fc2ba7 made ahrs_state 'volatile' in ahrs.c to prevent compiler optimization lockup 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1612 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-13 20:56:58 +00:00
peabody124
76971a807c AHRS: Some more tweaking of the timings. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1603 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-12 22:18:18 +00:00
peabody124
ba34bb8c24 AHRS: Oops, didn't notice the message processing was removed from end of loop. Adding back. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1597 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-12 12:44:57 +00:00
peabody124
59758122e1 AHRS: Cannot process messages while idling as this makes the main loop execution time non-deterministic based on OP (i.e. could get three messages and push the time up). It would be a nice featuer since you can get a raw and EKF update in the same step, but if the AHRS_Comm loop is fast (which is should be to keep update latency low) then this will hammer the AHRS with redunant requests. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1594 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-12 02:54:51 +00:00
peabody124
56131476ef AHRS: Switch to sampling at 2khz, plenty of CPU usage. However need to make EKF rate different for outdoor mode with this patch because otherwise will drop samples :-( 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1592 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-12 02:54:47 +00:00
stac
d1dc300081 ahrscomms: remove unused sync and switch_proto messages 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1582 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-11 23:06:25 +00:00
stac
80c92ba54b ahrs: process SPI messages while idle 
This makes the AHRS continually process SPI messages
while waiting for new samples to be ready.  This has
the side effect of making the CPU utilization on the
AHRS pin at 99-100% (ie. we make use of all of our
otherwise idle time).

git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1581 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-11 23:06:23 +00:00
peabody124
6cdb9b0b48 AHRS: Added Matlab code to analyze the INSGPS friendly data dump format and plot the important variables. This requires "DUMP_FRIENDLY" be enabled in the file. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1560 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-08 06:15:21 +00:00
peabody124
e499f33fd5 AHRS: Improved code to stream raw data from AHRS and added a matlab function to acquire and analyze the data. The QT program to analyze it should work but QextSerial seems to have a bug and drops or inserts data 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1559 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-08 06:15:18 +00:00
peabody124
9895b403c7 Ground/SerialLogging: Improved the serial dump from AHRS to have a big random frame header and to detect it on the logging program. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1556 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-07 16:33:10 +00:00
peabody124
e06162fa3a Flight/AHRS: Reorganized the code a little to make sure indoor used by default unless GPS updates come 
in.  Also change the velocity variance in indoor mode to lower to make it get a better attitude estimate (primarily faster convergence - probably will tweak this in the future dynamically).  Finally decreased barometer variance - in scope it's actually quite reasonable, normally within +/- .5 m.

git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1546 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-06 07:54:19 +00:00
peabody124
404e0c2f44 OP-132 AHRS: Changes to the Indoor flight handling to only use the barometer indoors and update with a zero for velocity and high variance. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1517 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-04 02:19:04 +00:00
peabody124
a406034ed9 OP-132 Flight/AHRS: Changed the heuristics regarding GPS quality and INSGPS updates. Now if satellites < 7 or PDOP > 3.5 it does not perform GPS updates. Professor is working on some code to fix parts of the covariance matrix. If indoor updates high a 0 position and velocity and high variance. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1515 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-09-04 02:19:00 +00:00
peabody124
12657440c3 OP-132 AHRS: Added an indoor flight mode and also add an ad-hoc quality heuristic to the GPS signal to increase the measurement variance from GPS when fix quality is poor or when in indoor flight mode. In addition corrected the indentation to conform to the tab/space convention forum discussion. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1474 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-31 02:11:09 +00:00
peabody124
f202884ce8 AHRS: Numerous changes. Added CPU load monitoring for AHRS as well as some friendly debugging output options. Also added a new CalibrationMode setting to AHRS (Echo) to speed up sensor calibration. Cleaned up the 6 point calibration interface to make it more friendly, but I think Edouard will clobber it with something nice and sexy :). 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1455 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-29 06:38:30 +00:00
peabody124
8128c3bbfa OP-118 AHRS: Added 6 point calibration routine to AHRS config widget. To use it, click "Cal Mode", then place the AHRS on each of 6 sides of a cube then click save position. Finally click compute to push the updated scale abd bias terms to the AHRS. Also fixed a bug in UAVObjectField that would not allow you to setDouble for extra fields. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1449 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-28 18:48:51 +00:00
stac
47df1faa8d ahrs: whitespace adjustments 
No functional changes.  Whitespace/tab fixups only.
Use TRUE/FALSE instead of 1/0.

git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1430 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-27 02:14:58 +00:00
stac
670452307d ahrs: remove unnecessary memset() of rx message buffer 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1429 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-27 02:14:56 +00:00
stac
0a7088c24a ahrs: fix type on flags, remove redundant code 
No need to have boolean flags being floats.
Some of the attitude message was being populated twice.

git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1428 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-27 02:14:54 +00:00
peabody124
c07e1416b7 AHRS: Serial output option for AHRS and corresponding serial logger. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1414 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-26 06:25:11 +00:00
peabody124
beac0c0ea9 AHRS: Get the signs for the simple algorithm correct 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1385 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-23 04:12:14 +00:00
peabody124
bd8fffe4d0 AHRS/Flight: Renamed AttitudeSettings to AhrsSettings to make it more intuitive. Added the ability to change the algorithm from GCS. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1383 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-23 03:39:13 +00:00
peabody124
2596f43511 OP-120 AHRS: Moved conversion of quaternion to RPY over from AHRS to OP to minimize transfer and computational load on AHRS. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1381 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-23 02:35:39 +00:00
peabody124
5662d85c4e OP-120 AHRS: Tightened up the communication protocol so it runs more quickly. Looks ok in gdb but GCS not crashing right now (rendering error) so I can't test it properly. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1366 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-22 00:21:54 +00:00
peabody124
8897c441fa OP-118 AHRS: Added AHRSCalibration object and messages associated with that to AHRS. Also made AHRS run calibration only when requested. We will create a GCS plugin that runs through computing the calibration settings. To compute the variances on the AHRS set the measure variances to true, wait ten seconds, then set it to false to read it back. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1363 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-21 21:46:02 +00:00
stac
b36d251bd9 ahrs: remove debug code in attitude handler 
Debug code was previously using the q4 element
in the attitude message to encode a cycle counter
to measure performance on the AHRS.  This is no
longer valid.

git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1357 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-21 16:23:14 +00:00
peabody124
2031cfe4d0 OP-119 AHRS: Added error counters for the messages as well as changed an error on AHRS echo back for the HomeLocation message. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1344 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-20 22:52:13 +00:00
peabody124
ef0b307353 [OP-119] Flight/AHRS: Created a HomeLocation object which is updated when the GPS gets an initial 3D fix and populated with the information to convert from LLA from the GPS to NED reference frame. Also added a message for passing the MagneticNorth vector to the AHRS and removed that computation from AHRS. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1337 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-19 20:38:27 +00:00
peabody124
3c4def13cc AHRS: Added message to pass GPS information into AHRS from OP. This is then converted 
into NED reference frame and used in the INSGPS algorithm, although currently this
information isn't propagated back to OP.  Data structures related to the GPS position
into the algorithm and the position estimate out will likely be in flux.

git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1334 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-19 05:18:34 +00:00
peabody124
cfd20922cd AHRS: Added the filtered data to the attitude raw object. Even though this seems like an oxymoron it's still raw to the EKF algorithm, and nice to see the real values after filtering to make sure nothing weird is going on. However, if people don't like this I'm open to suggestions/patches. I'd maybe suggest the whole attituderaw object be converted to ahrsraw anyway since it's more like that (eg. none of the values are actually attitude estimates). 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1325 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-18 14:54:24 +00:00
peabody124
4cba4929c2 AHRS: Fixed another bug in decimation code - I stupidly used Matlab 1 based indexing. Adjusted timing of AHRS so it can process all the data and doesn't miss updates. Finally adjust all inputs to INS to have correct sign and orientation (I think). This is X is forward, Y is right, and Z is down which is mentioned in the forum as our standard and fits the algorithm. The magnetic sensor driver should be modified to return in this order (currently reads X,Y,Z registers which maps to left, back, up) but maybe since it's a board dependent orientation it's better where it is. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1315 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-18 05:47:44 +00:00
peabody124
64bcebb87a AHRS: Fixed bug in the downsampling code (used a += a + blah, dumb). Also added initialization of the magnetic flux based on fixed GPS settings (I hope you're near Houston). I will add the GPS communication shortly. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1307 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-17 05:32:07 +00:00
peabody124
f74268b5e5 Cleanup and documentation. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1304 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-16 16:58:24 +00:00
peabody124
a23f3c66f4 Behaving reasonably, added calibration code which runs by default on startup and take 6 seconds - this will be called from GCS in future and saved to SD card. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1303 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-16 16:58:22 +00:00
peabody124
e0d63103ec Added a testing function for INS - currently doesn't stabilize for fixed inputs. Fixed two bugs I found in INSGPS (R[10] instead of R[9] initialized and Nav.Pos[2] was set to wrong X element). 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1302 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-16 16:58:20 +00:00
peabody124
d2e234b53a Adding the code from Dr. Dale to the AHRS. Currently diverges though, but not on the PC. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1301 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-16 16:58:18 +00:00
peabody124
ddbd4f8c80 Simple attitude calculation based on gravity and magnetic vector 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1298 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-15 06:03:44 +00:00
peabody124
3f408ad655 AHRS: Changing AHRS to use a DMA double buffering scheme with continuous samping for the analog data, then a simple fixed point FIR/decimate function. Currently all the oversampling parameters are hard coded as definitions, but will be updated to be configurable via settings. 
git-svn-id: svn://svn.openpilot.org/OpenPilot/trunk@1297 ebee16cc-31ac-478f-84a7-5cbb03baadba
 2010-08-15 04:09:01 +00:00