Commit Graph

964 Commits

Author SHA1 Message Date
Paul Riseborough 89be63d6c2 EKF: Fix vel pos innovation logging bug 2017-11-24 14:25:44 +11:00
Paul Riseborough 882f4d44e5
Merge pull request #362 from PX4/pr-ekfAuxVelFuse
EKF: Add additional velocity interface to use landing beacon data
2017-11-21 11:55:43 +11:00
Paul Riseborough b0ad8269a5 EKF: enable separate monitoring of aux velocity innovations 2017-11-21 11:18:09 +11:00
Paul Riseborough 2a57fd858d EKF: clean up reset of fusion flags 2017-11-21 11:18:09 +11:00
Paul Riseborough 8e30c2666d EKF: Add support auxiliary velocity observation
This enables the EKF to use an additional NE velocity measurement. This can be used to improve position hold stability when landing using a beacon system for positioning by fusing the beacon velocity estimates.
2017-11-21 11:18:09 +11:00
Paul Riseborough bd8f30752a
Merge pull request #351 from PX4/ekfGndEffectComp-wip
EKF: Protect against ground effect induced static pressure rise
2017-11-20 09:14:26 +11:00
Paul Riseborough a2b24fa960 EKF: Protect against ground effect induced static pressure rise during landing and takeoff.
Apply a dead-zone to the vertical position innovation if using baro for height and if in the ground effect region during and just after takeoff.
Method needs to be activated externally.
Turns off automatically after 10 seconds or if specified height gained.
2017-11-17 21:41:31 +11:00
Paul Riseborough c8d459eecd
Merge pull request #360 from dagar/pr-ekfObsBufAlloc
EKF RingBuffer allocation minor improvements + print memory usage
2017-11-17 08:38:57 +11:00
Paul Riseborough 3479542ae5
Merge pull request #359 from PX4/pr-ekfObsBufAlloc
EKF: Reduce memory usage
2017-11-17 08:09:37 +11:00
Daniel Agar d9c8af54bd
EKF add print_status() with memory usage 2017-11-16 14:11:34 -05:00
Daniel Agar 75d1ed894c
EKF simplify RingBuffer allocation check 2017-11-16 13:09:17 -05:00
Paul Riseborough af7004ef01 EKF: report observation buffer allocation errors 2017-11-16 20:04:46 +11:00
Paul Riseborough d785a19c0a EKF: delay final allocation of observation buffers until required 2017-11-16 19:47:04 +11:00
Paul Riseborough f54f1d467f
Merge pull request #358 from PX4/pr-ringbuffer_copies
EKF RingBuffer avoid unnecessary copying
2017-11-16 13:37:36 +11:00
Paul Riseborough 7c8fcf7628 EKF: Clarify use of *= operator for quaternions 2017-11-16 11:52:19 +11:00
Daniel Agar ed9a394029
EKF RingBuffer avoid copying 2017-11-15 18:08:51 -05:00
Paul Riseborough 44eaa076db
Merge pull request #356 from PX4/pr-ekfMiscUpdate
EKF: Remove redundant code
2017-11-16 07:05:18 +11:00
Paul Riseborough 2b20c52c4d EKF: Remove redundant code 2017-11-15 22:03:17 +11:00
Paul Riseborough cfdab732d1 EKF: Update parameter description 2017-11-15 21:28:11 +11:00
Paul Riseborough 16154423c8 matlab: update optical flow derivation
Support use of sensors with arbitrary orientation in body frame
2017-11-15 11:16:19 +11:00
Paul Riseborough cc651ac472
Merge pull request #354 from PX4/ekfMagYawReset-wip
Prevent loss of navigation accuracy after in-flight mag yaw alignment
2017-11-13 07:32:23 +11:00
Paul Riseborough 72a7ab2c25 EKF: Improve resistance to bad initial mag offset
When magnetic field states have been reset in-flight using a single sample, the magnetic field states are not used to constrain heading drift for a period after the reset. This period has been shortened from 10 to 5 seconds which is enough time to average out the effects of measurement noise (the original concern). The shorter time has enabled the previous practice for RW vehicles of using magnetic heading in that time period to constrain yaw drift to be discontinued. This is necessary becasue while magnetic heading is being used, it fights the yaw corrections obtained from GPs observations and lengthens the time required to recover from a bad mag calibration.
2017-11-13 07:05:56 +11:00
Paul Riseborough df9f48d2d3 EKF: Fix build error 2017-11-13 07:05:56 +11:00
Paul Riseborough 44c50ab2df EKF: Correct magnetic yaw measurement using learned mag biases
Don't apply bias corrections when biases are being learned to avoid possible circular data dependency.
2017-11-13 07:05:56 +11:00
Paul Riseborough c70363c501 EKF: Don't fuse heading if FW and waiting for mag states to stabilise
Doing so is a bad idea because bad mag data can drag the yaw angle away from the reset value and lead to rejection of GPS.
2017-11-13 07:05:56 +11:00
Paul Riseborough 279fc836f7 EKF: Always reset covariance matrix terms when doing vel pos state reset
If state errors were large before the reset, then failure to reset the covariance matrix terms can result in incorrect fusion of position and velocity measurements after the reset due to inconsistencies in the covariance matrix.
2017-11-13 07:05:56 +11:00
Paul Riseborough 141264fe63 EKF: Add method to set diagonals in covariance matrix 2017-11-13 07:05:56 +11:00
Paul Riseborough 32de90b9ef EKF: Add method to zero covariance terms 2017-11-13 07:05:56 +11:00
Paul Riseborough f3e34eddc9 EKF: do not attempt to align FW yaw using GPS method if on ground 2017-11-13 07:05:56 +11:00
Paul Riseborough 8f27d3fc54 EKF: don't reset quaternion states unnecessarily
When performing the initial in-flight mag yaw reset for RW vehicle, do not reset the quaternion states and corresponding variances unless there has been a change in yaw angle large enough to cause problems with navigation.
This is because the state estimates after a reset are more vulnerable to transient sensor errors, so a reset should be avoided if possible.
2017-11-13 07:05:56 +11:00
Paul Riseborough 9e47b6e1b6 EKF: don't reset quaternions unnecessarily
When performing the initial in-flight magnetic field reset for fixed wing vehicles, resetting the quaternion states and their corresponding covariances should be avoided unless yaw errors are large, because state estimates are vulnerable to transient sensor errors immediately following a reset.
2017-11-13 06:55:02 +11:00
Paul Riseborough 29d383edbf EKF: Allow mag field estimates to stabilise before use 2017-11-13 06:55:02 +11:00
Paul Riseborough 363edf5eb9 EKF: Fix yaw reset for fixed wing
Ensures that a complete reset of velocity and position states will always be performed if yaw has had to be reset using GPS velocity.
Ensures that the yaw_align status cannot be set to false once the filter has aligned.
2017-11-13 06:55:02 +11:00
Paul Riseborough 7852c0ed03
Merge pull request #312 from PX4/pr-ekfExtVisQuat
EKF: Enable operation with arbitrary External Vision reference frame
2017-11-12 08:35:39 +11:00
Paul Riseborough 5fd006ca48 EKF: remove redundant code 2017-11-11 07:06:01 +11:00
ChristophTobler 368248fe6d
Merge pull request #353 from PX4/pr-expo_int_cast
EKF: use uint64_t cast for XeY to avoid float casting of variables
2017-11-08 11:54:39 +01:00
ChristophTobler e388e59f32 EKF: use uint64_t cast for XeY to avoid float casting of variables
The default type for XeY is float
2017-11-08 11:40:26 +01:00
Paul Riseborough bba3f70a0e EKF: reduce prediction time step from 12 to 8 sec
Reduces susceptibility to incorrect estimation of acceleration bias during sustained yaw rate.
Requires an increase in RAM allocation of 837 Bytes to allow for the longer IMU and output predictor buffers that can be created.
2017-11-02 09:27:33 +01:00
Paul Riseborough 01d68ef67c EKF: Enable use of rotated external nav estimates 2017-11-01 08:33:57 +11:00
Paul Riseborough f921b2de5c
Merge pull request #348 from PX4/pr-ekfAddImuBiasReset
EKF: Add method to enable the IMU bias states to be reset externally
2017-10-30 15:55:52 +11:00
Paul Riseborough 063533afae EKF: Add method to enable the IMU bias states to be reset externally 2017-10-26 10:41:39 +11:00
Paul Riseborough cd2ca57ec2 Merge pull request #339 from PX4/ekfPosCtrlLimits-wip
Ekf pos ctrl limits wip
2017-10-23 06:57:23 +11:00
Daniel Agar 22a51c6afd PX4_ISFINITE -> ISFINITE 2017-10-22 20:37:40 +02:00
Daniel Agar 9e13a2cb21 tecs move initialization 2017-10-22 20:37:40 +02:00
Daniel Agar 31a8e047d7 tecs one line setters and group 2017-10-22 20:37:40 +02:00
Daniel Agar 4f2d571c89 tecs fix code style 2017-10-22 20:37:40 +02:00
Paul Riseborough 79995b2c15 Create total energy control system implementation
This is a new, clean and streamlined variant of the mathematical derivation I created a few years ago.
2017-10-22 20:37:40 +02:00
Daniel Agar 9eb860fabd travis-ci fix gcc and clang compilation jobs (#344)
- allow the python testing to fail without failing the entire build
2017-10-22 12:19:11 -04:00
Paul Riseborough e10ec59058 EKF: Use consistent test for navigation validity reporting
This will enable controller to take advantage of non-inertial dead reckoning.
2017-10-20 14:44:38 +11:00
Paul Riseborough 19074fdd9e EKF: Use consistent time limit for inertial dead reckoning test 2017-10-20 14:44:38 +11:00