Exceed Covariance Norm #830
Conversation
…e field to trigger reset, added ptest case and am waiting to receive values to write the state field to trigger the reset
.github/workflows/hootl.yml
Outdated
| @@ -56,6 +56,7 @@ jobs: | |||
| !(python -m ptest runsim -c ptest/configs/hootl_speedup.json -t FailingEmptyCase -ni) | |||
| !(python -m ptest runsim -c ptest/configs/hootl_speedup.json -t FailingEmptySimCase -ni) | |||
| python -m ptest runsim -c ptest/configs/hootl.json -t AttitudeNoSunVectorsInitializationCase -ni | |||
| python -m ptest runsim -c ptest/configs/hootl.json -t AttitudePersistanceExceededCase - ni | |||
| @@ -217,7 +221,7 @@ void AttitudeEstimator::_execute() | |||
| gnc::attitude_estimator_reset( | |||
| _state, time_s, {0.0f, 0.0f, 0.0f, 1.0f}); | |||
| } | |||
| else if (adcs_ssa_valid) | |||
| else if ((adcs_ssa_valid) || (exceed_persistance)) | |||
There was a problem hiding this comment.
Is this the logic we want? a comment explaining the logic/reasoning would be helpful here
There was a problem hiding this comment.
After a closer look at the code I realized I need to account for ssa valid vs invalid. In terms of what this does, I wanted to write something that would trigger attitude estimator reset if value is exceeded, but was unsure as to which reset it would be "gnc::attitude_estimator_reset(
_state, time_s, {0.0f, 0.0f, 0.0f, 1.0f});" or "gnc::attitude_estimator_reset(
_state, time_s, orbit_pos, b_body, adcs_ssa);"
| @@ -0,0 +1,855 @@ | |||
| #!/bin/bash | |||
There was a problem hiding this comment.
don't think we need this file
…baseline with safety factor
| @@ -43,6 +44,7 @@ AttitudeEstimator::AttitudeEstimator(StateFieldRegistry ®istry) | |||
| attitude_estimator_reset_cmd_f("attitude_estimator.reset_cmd", Serializer<bool>()), | |||
| attitude_estimator_mag_flag_f("attitude_estimator.mag_flag", Serializer<bool>(), 1), | |||
| attitude_estimator_ignore_sun_vectors_f("attitude_estimator.ignore_sun_vectors", Serializer<bool>(), 1), | |||
| attitude_estimator_reset_persistance_reached("attitude_estimator.reset_persistance_reached", Serializer<bool>(), 0), | |||
There was a problem hiding this comment.
why an eeprom save field value of zero here?
| /* The Current Frobenius Norm */ | ||
|
|
||
| float fro_norm= lin::fro(lin::nans<lin::Vector3d>()); | ||
|
|
||
| /* Make sure to set the attitude_estimator_reset_persistance_reached to false check this, this is the logic | ||
| that triggers the state field to allow for a reset of attitude_estimator. Once you exceed baseline frobenius norm | ||
| * 1000 (for the safety factor), trigger the persistance state field, accounting for SSA valid and invalid. */ | ||
|
|
||
| if (fro_norm > 3.86e-7 * 1000 && !adcs_ssa_valid) { | ||
| attitude_estimator_reset_persistance_reached.set(true); | ||
| } | ||
|
|
||
| else if (fro_norm > 8.35e-10 * 1000 && adcs_ssa_valid) { | ||
| attitude_estimator_reset_persistance_reached.set(true); | ||
| } | ||
|
|
||
| auto const exceed_persistance = attitude_estimator_reset_persistance_reached.get(); | ||
|
|
||
| /* One if statement for no ssa covariance and one if statement for ssa valid covariance */ | ||
|
|
||
| if (!adcs_ssa_valid && exceed_persistance) { | ||
| gnc::attitude_estimator_reset(_state, time_s, {0.0f, 0.0f, 0.0f, 1.0f}); | ||
| } | ||
| else if (adcs_ssa_valid && exceed_persistance) { | ||
| gnc::attitude_estimator_reset(_state, time_s, orbit_pos, b_body, adcs_ssa); | ||
| } | ||
|
|
* first commit :) * not working init flows * fixed error * small change * add adcs wheel1 faults * . * . * . Co-authored-by: Angela Loh <[email protected]>
…r-for-autonomous-navigation/FlightSoftware into exceed_covariance_norm
…e field to trigger reset, added ptest case and am waiting to receive values to write the state field to trigger the reset
…baseline with safety factor
|
We added persistence logic, created a writeable state field for the safety factor, and modified our ptest case to trigger the gnc::attitude_estimator_reset. The pictures above show the print statements we used to verify that a reset had occurred, and we tested by setting the safety factor extremely low to guarantee the triggering of exceed_persistance. |
|
We should close this PR as we unfortunately didn't have time to add this to the final FSW package. |
Modified Attitude_estimator ptest case and supporting files for a scenario that has been constructed (nothing written inside) for exceeding covariance norm. Added logic in attitude_estimator suite of files to trigger a reset when state field attitude_estimator_reset_persistance_reached becomes true (which triggers exceed_persistence), which could first be tested by manually triggering it in the ptest case but then adding the logic within attitude_estimator to trigger upon reaching some covariance.