Hi Camille,

Thanks a lot for the detailed reports. I just tested the files and can confirm the issue.

Well after a quick look in the scene you provided:

I'm not surprised the BarycentricMapping is not able to work .

Let's me explain:

• Object.DoFs is of type Rigid3d
• Object.Mechanical model.collisionDofs is of type Vec3d

What your are specifying by setting the input & output field of BarycentricMapping is that you are driving the collisionDofs(type Vec3d) from the one from the one of type Dofs(type Rigid3). So the corresponding mapping should be Rigid3d, Vec3d (an not Vec3d, Rigid3d).
The possible available templates shows that this is not supposed to work, because a BarycentricMapping can only have Vec3 as input. Overriding the template values with one that exists in BarycentricMapping will not work because this will lead to a mismatch between the input/output types and thes type manually set in the template.

Now about the BeamLinearMapping...the situation is a bit different. There is a valid template from Rigid3 to Vec3. This is why:

<BeamLinearMapping template="Rigid3d,Vec3d"/>

And

<BeamLinearMapping"/>

Work the same.
You get an error in your case because there is a typo "@collisionDOFs" instead of "@collisionDoFs".

Now I have no idea why the BeamLinearMapping is not working as you expect:

• maybe there is a real issue there ?

NB to anyone:
It would be nice to have PRs to improve the error messages as they are mis-leading.

Hi Damien,

Thanks a lot for your answer!

Regarding the BarycentricMapping, I indeed wasn't sure of which of the input and the output was driving the other (although I knew that the mechanical model was supposed to drive the collision model). It is definitely clear with your explanation. So I understand that there is no bug regarding the behaviour of the BarycentricMapping (as you pointed out), as it is not suppose to work for another input types than Vec3d.
For me, the (very minor) issue would rather be with the error message, as you said in your NB.

My apologies for the typo, indeed, I hadn't paid attention...
As for the unexpected behaviour, I haven't tried yet to track down what is happening. As I'm looking to use a reduced topology for my collision model, I might investigate it more in a near future. If so, I'll report here anything I find unsettling.

@Camille-K if I understand well, you want to make a beam structure (Rigid3d with its collision model) collide with a surface. Since the collision model is only instantiable in Vec3d you created a subnode with a mapping. Correct ?
@remibessard how do you manage this usually?

@hugtalbot Correct! This actually works with the IdentityMapping, but I'd be interested in using a reduced topology for the collision model.

are you coming to the STC @remibessard and @Camille-K ? it could be nice having you to work on this.
Otherwise this could be solved later

Unfortunately, I won't be able to come to Sophia Antipolis for the next STC... But if I haven't investigated the issue until then, I'll try to block 1 or 2 days during the STC to work on it from distance.
I'll keep you updated in the meantime.

Hi @remibessard
I am getting back to this topic.
I know students (Alban for instance) were struggling also with the BeamLinearMapping.
Can you update us on this topic ?
Is it working well for you?