Untested PriorPose3Ref #619
Conversation
| n_r = cf.factor.ref | ||
| nRp = p.x[2] | ||
| n_m = nRp * p_m | ||
| return n_r - n_m |
There was a problem hiding this comment.
Partials must be indicated, and for gravity should only be pitch and roll.
There was a problem hiding this comment.
It's not that simple. Pitch and Roll are not on the same manifold as the parameters or measurement.
I don't know yet how to make it work for the non-parametric solver, that's why no partials.
There was a problem hiding this comment.
Also, what if the sensor is mounted with a pitch of 90 deg?
There was a problem hiding this comment.
I think this is where the gradients come in - or info per coord
There was a problem hiding this comment.
that is why we are starting with gyro only, and we can get to accels once we are ready to work on them. This can become a rabbit hole of calib, 0vel detect etc. So I say just hold on accels entirely until we are ready for it. I need your help on the application examples please.
There was a problem hiding this comment.
I think results will be bad without absolute information. I've added partials as I see it.
There was a problem hiding this comment.
yep, agree adding gravity alignment will be good, let's just solve the bigger problem of alignments, rotations, and calibrations first. Will start using this factor soon after
Affie
force-pushed
the
22Q3/exp/Pose3Ref
branch
from
3739d29 to
3b99f9a
Compare
Codecov Report
@@ Coverage Diff @@
## master #619 +/- ##
==========================================
- Coverage 35.31% 35.20% -0.12%
==========================================
Files 48 48
Lines 1920 1926 +6
==========================================
Hits 678 678
- Misses 1242 1248 +6
📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
Idea for discussion.