[SOLVED] Error in RigidRigidMapping (v15.09)

download

Home › Forum › SOFA › Using SOFA › [SOLVED] Error in RigidRigidMapping (v15.09)

Viewing 9 posts - 1 through 9 (of 9 total)
  • Author
    Posts
  • 16 September 2015 at 13:29 #3596
    Deimel
    Blocked

    Hello all,

    I just noticed an error in the RigidRigidMapping when the parameter globalToLocalCoords is set to true. I created a test scene to show the problem.

    The upper node is without RigidRigidMapping (showing the positions of the DOFs).
    The lower node is with: <RigidRigidMapping repartition=”2 “globalToLocalCoords=”1” />

    The expected result is (imho) that the mapped DOFs (smaller trihedra) retain their global position, but some of them get displaced.

    git commit: a93d18d453b35b130abe1fc2dd52443549ca5ae4 (current v15.09 branch)

    Best,
    Raphael Deimel

    screenshot

    Scene Code

    16 September 2015 at 13:39 #3597
    Deimel
    Blocked

    Screenshot

    3 October 2015 at 09:39 #3730
    Deimel
    Blocked

    Seems like the rotations are applied to MechanicalObject’s in the wrong order. As long as either of the rotation property or the DoF’s orientation in the position property are not set (or set to 0 0 0 1), this bug does not affect the outcome.

    Here’s the simple fix:

    
diff --git a/modules/SofaBaseMechanics/MechanicalObject.cpp b/modules/SofaBaseMechanics/MechanicalObject.cpp
index 074ea8e..9c32cfe 100644
--- a/modules/SofaBaseMechanics/MechanicalObject.cpp
+++ b/modules/SofaBaseMechanics/MechanicalObject.cpp
@@ -132,7 +132,7 @@ void MechanicalObject<defaulttype::Rigid3dTypes>::applyRotation (const defaultty
     for (unsigned int i = 0; i < x.size(); i++)
     {
         x[i].getCenter() = q.rotate(x[i].getCenter());
-        x[i].getOrientation() *= q;
+        x[i].getOrientation() = q * x[i].getOrientation();
     }
 }

    It would be nice if somebody with repository access could fix the patch 🙂 Here’s a simple testcase to check the result:

    
<?xml version="1.0"?>

<!-- the point frame coordinates are rotated in global frame, but the orientation is rotated in local frame -->

<Node 	 name="root"  gravity="0 -9.81 0"  dt="1e-4"  time="0"  >
    <BackgroundSetting name="prettify" color=" 0 0 0" />
  	<ViewerSetting cameraMode="Orthographic" />
	<VisualStyle name="visualStyle1"  displayFlags="showVisual showBehaviorModels hideForceFields showInteractionForceFields hideCollision hideMapping hideOptions" />
	<MultiStepAnimationLoop collisionSteps="1" integrationSteps="5"/>
	
      <MechanicalObject template="Rigid3d" name="DOFs"  
            showObjectScale="2.5"  
            showObject="1" 
            rest_position="
0 0 0       0 0 0 1
32.1804178343 0.0 11.73187801      0.0 0.371991451914 0.0 -0.928236155137
42.4985667543 0.0 38.9022536239      0.0 0.702721776175 0.0 -0.711464760399
		    "
    		translation="0 0 0" 
    		rotation="0 0 0"
		/>
	
      <MechanicalObject template="Rigid3d" name="DOFs"  
            showObjectScale="2.5"  
            showObject="1" 
            rest_position="
0 0 0       0 0 0 1
32.1804178343 0.0 11.73187801      0.0 0.371991451914 0.0 -0.928236155137
42.4985667543 0.0 38.9022536239      0.0 0.702721776175 0.0 -0.711464760399
		    "
    		translation="0 0 0" 
    		rotation="0 0 30"
		/>
 
	</Node>
</Node>
    3 October 2015 at 09:44 #3731
    Deimel
    Blocked

    The forum does not seem to handle code correctly :/

    Patch:
    http://pastebin.com/cE3X4YEN

    Testcase:
    http://pastebin.com/X10WnYPH

    4 October 2015 at 11:00 #3734
    Hugo
    Keymaster

    Thank you very much Raphael for your post ! This information will be forwarded to the developers.

    By the way, concerning the forum, it is now possible to:

    
insert code using the 
     instruction before your code

    

    but we acknowledge it is not perfect yet !
    

    The format of the users' profile has been corrected : the nickname will really be used in the forum. You just need to re-edit the profile once and fill in the nickname.
    

    Cheers,
    

    Hugo
    6 October 2015 at 13:19 #3740
    max
    Blocked

    Hello Raphael,

    I had a quick look at your test scene and it seems that the repartition is the issue.

    More precisely, if you get rid of the repartition field in the mapping, like this:

    <RigidRigidMapping globalToLocalCoords="1" />

    then the problem seems to go away. Can you confirm this ?

    The semantic for repartition is somewhat complex, and admittedly the doc does not help much. I did not dwelve into it yet (anyone?) but as long as you simply want 1->1 mappings between rigid bodies, you’d better remove repartition and index informations altogether.

    Hope this helps,

    Maxime

    13 October 2015 at 22:42 #3743
    Deimel
    Blocked

    Hello Maxime,

    Please have a look at the testcase I posted, it doesn’t contain a RigidRigidMapping directive but nevertheless rotates the DOFs around the wrong axis (the DoF’s frame instead of the global frame).

    Best,
    Raphael

    21 October 2015 at 16:46 #3811
    max
    Blocked

    Hello Deimel, and sorry for the delay.

    You are correct, the multiplication order is inconsistent.

    2c9f6b4 on master applies your changes (+ float rigid types). Do we need to backport it to another branch ?

    Thanks again for your patch.

    Best,

    max.

    24 October 2015 at 19:00 #3872
    Deimel
    Blocked

    It would probably be nice to also fix it in the current “stable” v15-09 branch.

    Best,
    Raphael

  • Author
    Posts
Viewing 9 posts - 1 through 9 (of 9 total)
  • You must be logged in to reply to this topic.
Back to Top