Models interpenetrating each other after collision

[ZacharyZhou12138]

Hi :)

I have created a scene of simulating the oral stage of swallowing(motors not included, only physical actuators are simulated):

from tongue import Tongue
from rigidObject import Oralpassage
from zhoufar import Zhoufar
from zhouclose import Zhouclose
from larynx import Larynx
from LarynxController import MultiLarynxController

def createScene(rootNode):

    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Constraint.Lagrangian.Correction')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Engine.Select')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.LinearSolver.Direct')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Mapping.Linear')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Mass')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.ODESolver.Backward')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.SolidMechanics.FEM.Elastic')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.SolidMechanics.Spring')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.StateContainer')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Topology.Container.Dynamic')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Visual')       
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.AnimationLoop')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Collision.Detection.Algorithm')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Collision.Detection.Intersection')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Collision.Geometry')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Collision.Response.Contact')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Constraint.Lagrangian.Solver')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.LinearSolver.Iterative')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Mapping.NonLinear')  
    rootNode.addObject('RequiredPlugin', name='Sofa.Component.Topology.Container.Constant')  
    rootNode.addObject('RequiredPlugin', name='MultiThreading')

    
    from stlib3.scene import MainHeader, ContactHeader
    MainHeader(rootNode, gravity=[0.0, 0.0, -981.0], 
               plugins=["SoftRobots", "SofaPython3", "Sofa.Component.AnimationLoop",
                        "Sofa.Component.Collision.Detection.Algorithm",
                        "Sofa.Component.Collision.Detection.Intersection",
                        "Sofa.Component.Collision.Geometry",
                        "Sofa.Component.Collision.Response.Contact",
                        "Sofa.Component.Constraint.Lagrangian.Correction",
                        "Sofa.Component.Constraint.Lagrangian.Solver",
                        "Sofa.Component.IO.Mesh",
                        "Sofa.Component.LinearSolver.Iterative",
                        "Sofa.Component.Mass",
                        "Sofa.Component.ODESolver.Backward",
                        "Sofa.Component.StateContainer",
                        "Sofa.Component.Topology.Container.Constant",
                        "Sofa.Component.Visual",
                        "Sofa.GL.Component.Rendering3D"])

    ContactHeader(rootNode, alarmDistance=3, contactDistance=1.4, frictionCoef=0.08)
    rootNode.VisualStyle.displayFlags = "showForceFields showBehaviorModels showInteractionForceFields"

    rootNode.addObject('CGLinearSolver', 
                         template="CompressedRowSparseMatrixMat3x3d", 
                         iterations=25,   
                         tolerance=1e-9)   

    rootNode.addObject('AsyncSparseLDLSolver', template="CompressedRowSparseMatrixMat3x3d")

    freeMotion = rootNode.getObject('FreeMotionAnimationLoop')

    if freeMotion is not None:
        freeMotion.parallelODESolving.value = True
        freeMotion.parallelCollisionDetectionAndFreeMotion.value = True

    zhoufar = Zhoufar(rootNode)
    zhouclose = Zhouclose(rootNode)
    larynx = Larynx(rootNode)

    controller = MultiLarynxController(
        rigidNodes=[zhoufar, zhouclose, larynx],
        speeds=[30.0, 30.0, 30.0],
        step_distances=[27.0, 20.0, 16.0],
        delays=[0.0, 0.5, 0.8]
    )
    rootNode.addObject(controller)

    Oralpassage(rootNode, rotation=[0, 0, 0], translation=[0, 0, 0],
                color=[0, 0.749, 1, 0.4], isAStaticObject=True)

    Tongue(rootNode, name="Tongue", rotation=[0, 0, 0], translation=[0, 0, 0])

    return rootNode

After collision, the tongue interpenetrate the oral passage. I know that with contactDistance being set too low, the models may interpenetrate eachother. However, if I set the contactDistance even slightly higher (like1.5), the tongue will just disappear at the beginning of the simulation and sofa crashes at the start of the simulation (the simulation stops, and the software becomes unresponsive(I think it's because the distance between the tongue and the second elevating actuator is already shorter than 1.5).

Simulation result when contactDistance=1.4 :

testorgans__r60_0026.mp4

[bakpaul]

Hello !

Because you are using functions to generate your scene, looking at your source code doesn't provide any information to me, I'll only answer given your comments and the video.

This threshold you are facing might come from the fact that at rest your object already has a distance bellow 1.5 to an obstacle. Meaning that at the first timestep, a force need to be applied to push it. This might indeed break the scene because the required force might be too large.

I don't know which primitive you are using for the contact. It is better to always use triangles (or quads) for one and lines or points for the other one (depending which one works best). Is the blue part static (meaning not deformable and never moving ?). If so you can specify on its collision model moving=False, simulated=False. Specifying this will help both collision detection and solving. If so, I suggest using the triangles on the static part.

Another point is to make sure your GenericConstraintSolver is converging (you can look at its tab 'stats' and if the number of steps is bellow the max iteration data you provided then you've converged). You have two solutions following this :

1. you are converging, this might mean the the error threshold you provided is too high to guarantee no interpenetration
2. you are not converging, so you need to increase the number of iteration to let it converge

Finally : To insure no volume interpenetration you can use a plugin that uses distance grid for collision. It is called DistanceGrid plugin. s you model is loaded from a file you can directly use this file to generate the distance grid and use this for collision detection. This works best if the object with the distance grid is not moving, so I suggest using it on the blue part. See this scene for an example.

Hope that helped !

[github-actions[bot]]

⚠️ ⚠️ ⚠️
@ZacharyZhou12138
Feedback has been given to you by the project reviewers, however we have not received a response from you. Without further news in the coming weeks, this discussion will be automatically closed in order to keep this forum clean and fresh 🌱 Thank you for your understanding

[github-actions[bot]]

⚠️ ⚠️ ⚠️
@ZacharyZhou12138
In accordance with our forum management policy, the last reply is more than 4 months old and is therefore closed. Our objective is to keep the forum up to date and offer the best support experience.

Please feel free to reopen it if the topic is still active by providing us with an update. Please feel free to open a new thread at any time - we'll be happy to help where and when we can.