Fix coordinate frames on state message #81
Conversation
chapulina
commented
Nov 16, 2021
•
chapulina
commented
- Wrap up Document / update coordinates and units #48
- Closes LRAUVState fields population status #11
- Axis visualization: ENU, NED, FSK #104 helps understanding this PR, but it's not required
- Make sure all fields are documented
- Test all populated fields on state message except for sensor data (wait for interpolation)
- Update published message to make sure quantities are in the documented frames.
Signed-off-by: Louise Poubel <[email protected]>
Signed-off-by: Louise Poubel <[email protected]>
chapulina
force-pushed
the
chapulina/more_docs
branch
from
41ccccf
to
0f8c4be
Compare
Signed-off-by: Louise Poubel <[email protected]>
Signed-off-by: Louise Poubel <[email protected]>
Signed-off-by: Louise Poubel <[email protected]>
chapulina
changed the title
Signed-off-by: Louise Poubel <[email protected]>
|
Hey, @chapulina I've not had time to debug the failures in this PR. But on my PC when I run the tests I'm also seeing segfaulting and regression in a lot of the integration test cases. It seems to cause the depth_pitchmass_vbs test to segfault consistently... |
Signed-off-by: Louise Poubel <[email protected]>
chapulina
force-pushed
the
chapulina/more_docs
branch
from
d4f53b0
to
512bd2f
Compare
|
I narrowed down the controller failures to the RPH orientation. I'm still trying to understand the expectations around those values, see comments on 5903584. |
…reference frame Signed-off-by: Louise Poubel <[email protected]>
chapulina
force-pushed
the
chapulina/more_docs
branch
from
512bd2f
to
5903584
Compare
| // | ||
| // [VerticalControl](CRITICAL): Excessive depth excursion=10.573917 m, failToGoUpDepth_=17.503380 m, depthRate=0.211946 m/s, pitch =0.114867 deg. | ||
| // | ||
| // The difference between poseOffsetNED.Rot() and rph is that roll and pitch are swapped. |
There was a problem hiding this comment.
Here's a simple example, I manually pitched the vehicle's nose down. The frames shown are ENU (red: East, green: North, blue: Up).
This corresponds to a negative pitch about North (green). That's the first element of NED, and the second of END. The print outs of the line below are consistent with that:
NED: -0.505718 0.000121 0 -- END: 0.000138 -0.505718 6.8e-05
I think that what could be happening is that the controller expects the vehicle to start facing North instead of West. A nosedown pitch with the robot facing North would be negative on East, and NED would give us a negative angle on it's Y coordinate.
There was a problem hiding this comment.
I just verified that if the vehicle starts facing North, the NED orientation works for the controllers.
There was a problem hiding this comment.
diff --git a/lrauv_ignition_plugins/worlds/buoyant_tethys.sdf b/lrauv_ignition_plugins/worlds/buoyant_tethys.sdf
index 1618c1e..412ffbb 100644
--- a/lrauv_ignition_plugins/worlds/buoyant_tethys.sdf
+++ b/lrauv_ignition_plugins/worlds/buoyant_tethys.sdf
@@ -315,7 +315,7 @@
</include-->
<include>
- <pose>0 0 0 0 0 0</pose>
+ <pose degrees="true">0 0 0 0 0 -90</pose>
<uri>tethys_equipped</uri>
</include>… expectations Signed-off-by: Louise Poubel <[email protected]>
…ace North Signed-off-by: Louise Poubel <[email protected]>
|
Moving this PR to draft. I'll be breaking some of the smaller changes into new PRs, and update this PR to reorient the model in a way that it faces North at zero orientation. While at it, I'll also tackle #80 . |
Signed-off-by: Louise Poubel <[email protected]>
chapulina
changed the title
Signed-off-by: Louise Poubel <[email protected]>
Signed-off-by: Louise Poubel <[email protected]>
…ctly Signed-off-by: Louise Poubel <[email protected]>
|
Closing this PR, it's being broken into smaller ones, starting with: |
