Commit Graph

722 Commits

Author SHA1 Message Date
Daniel Agar dc8ed97809 ekf2: optical flow control limits constrain speed using HAGL max 2022-06-21 12:52:25 -04:00
Daniel Agar d5839e2dd5 optical flow sensor pipeline overhaul
- all sources of optical flow publish sensor_optical_flow
 - sensor_optical_flow is aggregated by the sensors module, aligned with integrated gyro, and published as vehicle_optical_flow

Co-authored-by: alexklimaj <alex@arkelectron.com>
2022-06-20 20:56:56 -04:00
Daniel Agar b81a5b3efa ekf2: request mag yaw reset after calibration or sensor change 2022-06-17 21:04:05 -04:00
Daniel Agar 1980b5c5e8 ekf2: setEkfGlobalOrigin() reset baro and hgt sensor offsets if necessary
- handle uninitalized _gps_alt_ref
 - add basic lat/lon/alt sanity checks
2022-06-16 00:59:54 -04:00
bresch 0d256b8ff6 ekf2 wind: use noise spectral density for process noise tuning
The noise spectral density, NSD, (square root of power spectral density) is a
continuous-time parameter that makes the tuning independent from the EKF
prediction rate.
NSD corresponds to the rate at which the state uncertainty increases
when no measurements are fused into the filter.
Given that the current prediction rate of EKF2 is 100Hz, the
same tuning is obtained by dividing the std_dev legacy parameter by 10:
nsd = sqrt(std_dev^2 / 100Hz)
2022-06-14 18:39:10 +10:00
bresch 4f1091792f ekf2 preflight: only check innovation of active height sources 2022-06-07 11:44:56 -04:00
PX4BuildBot a247b42907 [AUTO COMMIT] update change indication 2022-06-04 15:57:11 -04:00
bresch e2955bdd61 terrain est: clear innovation/var/test ration when aiding stops 2022-06-03 17:31:14 -04:00
Daniel Agar 21b1c933dc ekf2: EKFGSF_yaw delete unnecessary internal state 2022-06-01 13:16:06 -04:00
Daniel Agar 3889b79342 ekf2: yaw estimator add yaw_composite_valid boolean 2022-06-01 13:16:06 -04:00
Daniel Agar 095e0f0604 ekf2: sensor_simulator fix enum shadow 2022-05-29 00:21:08 -04:00
Daniel Agar 58ea6235fe ekf2: estimator aid source status (range height) 2022-05-25 09:25:12 -04:00
Daniel Agar 1e25aee6fa ekf2: estimator aid source status (baro height) 2022-05-25 09:25:12 -04:00
Daniel Agar 639222dd65 ekf2: estimator aid source status (GPS pos, GPS vel) 2022-05-23 16:21:49 -04:00
Daniel Agar 1ae467e9cd ekf2: estimator aid source status (starting with fake position) 2022-05-23 16:21:49 -04:00
bresch c57c575cfe ekf2: use explicit flags instead of bitmask position
This prevents bitmask mismatch when a new flag is inserted
2022-05-23 14:43:02 -04:00
bresch 00e88a02c0 baro bias: improve innovation sequence monitoring
The baro observation noise parameter is often over-estimated in order as
a measure to mitigate temporary offsets in the readings due to wind
gusts or poor pressure compensation tuning. The side effect is that the
innovation sequence monitoring based on normalized innovation struggles
to detect an offset in the state because the innovation isn't
statistically significant.
To counter this issue, a simpler check is added to trigger the process
noise boost when the innovation has the same sign for a long period of
time.
2022-05-21 22:37:04 +02:00
Daniel Agar db4545e442 Update world_magnetic_model to latest Sat May 21 11:14:10 UTC 2022 2022-05-21 14:57:55 -04:00
mcsauder ea10eacb99 Replace EKF/common.h #defines with enums. 2022-05-18 09:25:19 -04:00
mcsauder 2a66be4899 Colocate struct stateSample with other instances of *Sample structs. 2022-05-05 18:41:50 -04:00
mcsauder a0d9687409 Add c++ style initializers where missing in EKF/common.cpp, standardize tab/space indentation, align comments and format whitespace. 2022-05-05 18:41:50 -04:00
bresch cab477d715 ekf2: optimize KHP computation
Calculating K(HP) takes less operations than (KH)P because K and H are
vectors.

Without considering the sparsity optimization:
- KH (24*24 operations) is then a 24x24 matrix an it takes
24^3 operations to multiply it with P. Total: 14400 op

- HP (24*(24+24-1) operations) is a row vector
and it takes 24 operations to left-multiply it by K. Total:1152 op
2022-05-05 10:34:17 -04:00
Daniel Agar 6fc857772d ekf2: remove unnecessary const references 2022-04-29 21:39:02 -04:00
Daniel Agar d1d15a6f6d differential pressure remove filters from drivers and average in sensors/airspeed 2022-04-26 08:39:19 -04:00
bresch e32a484906 [AUTO COMMIT] update change indication 2022-04-25 12:58:18 -04:00
bresch db0274e19b ekf: robustify bad_acc_vertical check
when the vertical pos or vel innov ratio is above the threshold, the
other one needs to be significant too and not just positive to trigger
the failure
2022-04-25 12:58:18 -04:00
bresch e9c07fac6f EKF: move python tuning tools to EKF module 2022-04-21 13:39:26 +02:00
bresch 37cafe7dcd ekf rng kin: reduce minimum rng variance 2022-04-13 11:41:48 +02:00
bresch 4994649500 ekf rng kin: increase default gate size
The user needs to tune the range finder noise parameters properly and we
shouldn't need such a small gate
2022-04-13 11:41:48 +02:00
bresch 1fbe04986f ekf rng finder consistency: simplify class member names 2022-04-13 11:41:48 +02:00
bresch 4c03f0bc50 ekf: make range finder kin consistency gate tunable by parameter 2022-04-13 11:41:48 +02:00
bresch 079a5e92ba ekf: run rng consistency check only when not horizontally moving
The check assumes a non-moving terrain height
2022-04-13 11:41:48 +02:00
bresch d903613c9c ekf: add logging for rng kinematic consistency check 2022-04-13 11:41:48 +02:00
bresch 8693ad15a7 ekf: add logging of range finder consistency check 2022-04-13 11:41:48 +02:00
bresch baf9cc9597 ekf: use uint64_t for time variables 2022-04-13 11:41:48 +02:00
bresch 9fc331b7ea ekf: requires kinematically consistent range finder data to continue terrain aiding 2022-04-13 11:41:48 +02:00
bresch 78211f9dbb ekf: improve rng consistency check
To pass from invalid to valid:
- time hysteresis
- some vertical velocity
- test ratio < 1
- low-passed signed test ratio < 1

To pass from valid to invalid:
- low-passed signed test ratio >= 1
2022-04-13 11:41:48 +02:00
bresch b1ea2e4e15 ekf: use same gate for innov and innov sequence monitoring 2022-04-13 11:41:48 +02:00
bresch f96287b80a ekf: access member variable without getter 2022-04-13 11:41:48 +02:00
bresch 904bf8ef9f ekf: add range finder kinematic consistency check
At each new valid range measurement, the time derivative of the distance
to the ground is computed and compared with the estimated velocity.
The average of a normalized innovation squared statistic check is used
to detect a bias in the derivative of distance measurement,
indicating that the distance measurements are kinematically inconsistent
with the filter.
2022-04-13 11:41:48 +02:00
bresch 064518f57a ekf: extract range finder noise computation 2022-04-13 11:41:48 +02:00
Daniel Agar 8166a500ac Update world_magnetic_model to latest Mon Apr 11 11:14:11 UTC 2022 (#19475) 2022-04-12 10:11:51 -04:00
Daniel Agar bb2ea574aa ekf2: properly reset IMU biases on calibration change (non-multi-EKF)
- this was working in the multi-EKF case using vehicle_imu, but missing
in sensor_combined
2022-04-11 12:23:55 -04:00
Daniel Agar 04f37222f8 ekf2: fix IMU missed perf count when not using multi-EKF 2022-04-10 11:07:33 -04:00
Daniel Agar 0c31f63896 sensors: add baro calibration and cleanup
- sensor_baro.msg use SI (pressure in Pascals)
 - update all barometer drivers to publish directly and remove PX4Barometer helper
 - introduce baro cal (offset) mainly as a mechanism to adjust
relative priority
 - commander: add simple baro cal that sets baro offsets to align with
GPS altitude (if available)
 - create new sensors_status.msg to generalize sensor reporting
2022-03-26 16:08:41 -04:00
Daniel Agar f4c3084c26 init adjustments to ensure used topics are advertised early (primarily for logging)
- multi-EKF create each instance as soon as IMU & mag are advertised (before device id populated)
2022-03-23 16:51:16 -04:00
Daniel Agar 7ef38112d2 ekf2: return saved mag bias variance when not in 3d magnetometer fusion
- the estimated mag bias was requiring > 30 seconds of continuous 3d
mag fusion to be reported stable (and saved back to mag cal), this
restores the original intent requiring 30 seconds of accumulated valid
3d mag fusion
2022-03-14 21:03:41 -04:00
Daniel Agar 5717434e93 ekf2: initialize GPS drift metrics to NAN rather than 0 2022-03-14 10:09:24 -04:00
Daniel Agar a430f0ccae ekf2: add simple zero velocity update when vehicle at rest (#19149)
- further decreases initial tilt alignment time (now down to 2.5 seconds if still) and improves initial bias estimates
2022-03-12 12:56:31 -05:00
wangwwno1 cd5a1e510a ekf2: typo Fix: pub.advertised() -> pub.advertise() (#19302) 2022-03-09 08:39:08 +01:00