Trouble using AttachConstraint

[ISIR-GuillaumeS]

Hi guys !

I have some troubles using the node.addobject("AttachConstraint") function.

To begin with, I did not fully understand how to obtain the list of indices of the two objects that we want to attach. For instance, in the example "AttachConstraint.scn" (that you can find here sofa/src/examples/Component/Constraint/Projective) we can find :

<AttachConstraint object1="@M1" object2="@M2" indices1="144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159" indices2="0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15" constraintFactor="1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1"/>

Problem is that how indices1 and indices2 are obtained ?
The second thing is that I tried to implement that part in my case (and using python). I would like to constrain 2 disks (which are mechanical objects) defined below :

def Disque(solverNode, translation, rotation):
    #### MAPPING of the disks (here some torus) ####    ###########
    disque = solverNode.cosserat.rigidBase.cosseratInSofaFrameNode.addChild("DisqueMechanics")

    totalMass = 0.006
    volume = 0.000004247    #diametre = 52; epaisseur=2; masse=6.35g
    inertiaMatrix = [0.000002028, 0.0, 0.0, 0.0, 0.000002028, 0.0, 0.0, 0.0, 0.000004056]

    disque.addObject('EulerImplicit', name='odesolver')
    disque.addObject('CGLinearSolver', name='Solver')
    

    disque.addObject('MechanicalObject', name="DOF", template="Rigid3", translation=translation, rotation=rotation)
    disque.addObject('UniformMass', name="vertexMass", vertexMass=[totalMass, volume, inertiaMatrix[:]])

    # disque.addObject('UncoupledConstraintCorrection')

    mapping = disque.addChild("Mapping")
    mapping.addObject("MeshOBJLoader", name="meshLoader_2", filename="Disque.obj", scale="0.001", handleSeams="1", scale3d="1 1 1")
    mapping.addObject("OglModel", name="Visual", src="@meshLoader_2", color="1.0 0.5 0.25 1.0")
    mapping.addObject("RigidMapping", input="@..", output="@Visual", index="10")

    ###########    ###########    ###########    ###########
    return disque

Finally, in my creatScene function I added (I chose indices randomly):
rootNode.addObject('AttachConstraint', object1="@d1", object2="@D2", indices1="1", indices2="11", constraintFactor="1")

However, this gave me this error :

[ERROR]   [SofaPython3::SceneLoader] Unable to completely load the scene from file '/home/benamar/Documents/Travail/SimuSofa/PCS_Example1.py'.  
Python exception:   
  ValueError: Object type AttachConstraint<> was not created  
The object is in the factory but cannot be created.  
Requested template : None  
Used template      : None  
Also tried to create the object with the template 'Rigid2d' but failed for the following reason(s):  
    1. Data attribute 'object1' does not point to a valid mechanical state of datatype 'Rigid2d'.  
Also tried to create the object with the template 'Rigid3d' but failed for the following reason(s):  
    1. Data attribute 'object1' does not point to a valid mechanical state of datatype 'Rigid3d'.  
Also tried to create the object with the template 'Vec1d' but failed for the following reason(s):  
    1. Data attribute 'object1' does not point to a valid mechanical state of datatype 'Vec1d'.  
Also tried to create the object with the template 'Vec2d' but failed for the following reason(s):  
    1. Data attribute 'object1' does not point to a valid mechanical state of datatype 'Vec2d'.  
Also tried to create the object with the template 'Vec3d' but failed for the following reason(s):  
    1. Data attribute 'object1' does not point to a valid mechanical state of datatype 'Vec3d'.  

Do you have an idea ?

[hugtalbot]

Hi @ISIR-GuillaumeS

For your first question: The selection of indices is still a pain in SOFA, you can find two tuto here:

• Micro tuto SOFA: custom node selection
• Micro tuto SOFA: visualization of point indices

For your second question: AttachConstraint is a projective constraint which couples the two objects to have some of their degrees of freedom equal at the end of each time step through a projection. This is a "weak" projective constraint. It therefore assumes that both MechanicalObject that you couple have the same template.

[ISIR-GuillaumeS]

@hugtalbot I tried to do the same thing as the first tuto but it still doesn't work. I added at the end of the Disque function :

disque.addObject('MeshOBJLoader', name="IdLoader", filename="Points2.obj")
disque.addObject('FixedConstraint', indices='@../IdLoader.position')

With Points2.obj that correspond to the csv file here
Points2 (copie).csv

When I try it, I only see one point appearing at the origin of axis (not even on my disk).
However, when I replace indices='@../IdLoader.position' by some ID that I found in the CSV file, for example :

disque.addObject('MeshOBJLoader', name="IdLoader", filename="Points2.obj")
disque.addObject('FixedConstraint', indices='1072 1237 1460 1723 1748 1754 1876 1903')

No points are appearing (not even the one at the origin) but I have a warnings telling :

[WARNING] [FixedConstraint(FixedConstraint)] Index 1072 not valid, should be [0,1]. Constraint will be removed.
[WARNING] [FixedConstraint(FixedConstraint)] Index 1237 not valid, should be [0,1]. Constraint will be removed.
[WARNING] [FixedConstraint(FixedConstraint)] Index 1460 not valid, should be [0,1]. Constraint will be removed.
[WARNING] [FixedConstraint(FixedConstraint)] Index 1723 not valid, should be [0,1]. Constraint will be removed.
[WARNING] [FixedConstraint(FixedConstraint)] Index 1748 not valid, should be [0,1]. Constraint will be removed.
[WARNING] [FixedConstraint(FixedConstraint)] Index 1754 not valid, should be [0,1]. Constraint will be removed.
[WARNING] [FixedConstraint(FixedConstraint)] Index 1876 not valid, should be [0,1]. Constraint will be removed.
[WARNING] [FixedConstraint(FixedConstraint)] Index 1903 not valid, should be [0,1]. Constraint will be removed.

Do you have an idea why ?

[hugtalbot]

Any update @ISIR-GuillaumeS ?
Have you solved your issue?

[ISIR-GuillaumeS]

No, not really. I adopted another solution to conduct the scene I wanted but thanks though !

[hugtalbot]

An alternative could be Lagrangian constraint to control dofs, like in #4512