ContactListener API between rigid body and soft body

[thundermms]

Problem

Description
I am very new to sofa and still trying to understand and navigate my way around it, so sorry if my questions seems to be trivial.
I am trying to find the point of contact between two collision models ( in this case a gripper (rigid body) and a soft body layer. I decided to use ContactListener API. Now I make the gripper drop from up at a controlled speed, it touches once and bounces back. I get two sets of data for this scene, as attached below.
example output:
{'numberOfContacts': 16, 'collisionPointsModel1': [Vec3d(0, 2.76162, 0.013304), Vec3d(0, 2.7614, 0.0137588), Vec3d(0.00186864, 2.76889, 0.0547318), Vec3d(0.0305556, 2.7647, 0.0445549), Vec3d(0.0382823, 2.7616, 0.0308967), Vec3d(0.0508172, 2.76565, 0.00901991), Vec3d(0.046474, 2.76377, 0.0204908), Vec3d(0.0488622, 2.76533, 0.0215185), Vec3d(0.0454395, 2.76252, 0.0158121), Vec3d(-0.0179506, 2.78136, 0.068998), Vec3d(0.0393523, 2.75932, -0.0142727), Vec3d(0.0358865, 2.77477, -0.0556169), Vec3d(0.0377225, 2.75761, -0.0056905), Vec3d(0.0637444, 2.78346, -0.0322426), Vec3d(0.0639254, 2.77955, -0.0166332), Vec3d(0.0692672, 2.79861, -0.0434354)], 'collisionPointsModel2': [Vec3d(0.0526529, 2.70831, 0.0617356), Vec3d(0.052426, 2.70831, 0.0619818), Vec3d(0.032873, 2.70827, 0.0831928), Vec3d(0.051119, 2.70831, 0.0633995), Vec3d(0.0614721, 2.70833, 0.0521685), Vec3d(0.0781368, 2.70837, 0.0340907), Vec3d(0.0704232, 2.70835, 0.0424584), Vec3d(0.0710097, 2.70835, 0.0418221), Vec3d(0.0722783, 2.70836, 0.040446), Vec3d(0.016671, 2.70824, 0.100769), Vec3d(0.0844714, 2.70838, 0.0272188), Vec3d(0.103498, 2.70842, 0.00657852), Vec3d(0.0794441, 2.70837, 0.0326724), Vec3d(0.104656, 2.70843, 0.00532265), Vec3d(0.0969563, 2.70841, 0.0136752), Vec3d(0.112786, 2.70844, -0.0034967)], 'collisionElementsModel1': [5187, 9129, 5984, 5845, 5940, 5827, 5934, 5933, 5830, 9527, 5899, 5605, 5800, 5641, 5756, 5612], 'collisionElementsModel2': [304, 304, 304, 304, 304, 304, 304, 304, 304, 304, 304, 304, 304, 304, 304, 304]}
Now my questions are

1. The collisionPointsModel1 and collisionPointsModel2 are in vec3D and are supposed to give the coordinates of the points of contact, are these in cartisean frame, and what is the origin of this ?
2. collisionElementsModel1 and collisionElementsModel2 are integer numbers, I initially thought these are the elements of the Mechanicalobject of each module, but that was not the case as the gripper is a rigid body and I should have only received 1 value, but I get a lot more values as shown in the output. Can I know what elements these are referring to if not Mechanical Object
3. Currently I have used LineCollisionModel in this line, I also have TriangleCollisionModel, PointCollisionModel and TetrahedronCollisionModel along with LineCollisionModel. May I know what would be the difference in each and what would be the right model to use ?
   root.addObject("ContactListener", name="collisionListener", collisionModel1= probe_node.probeCollision.LineCollisionModel.getLinkPath(), collisionModel2= skin.skinCollision.LineCollisionModel.getLinkPath())
4. I basically want to find the point of force contacts and their forces, am I on the right track or is there another way to find the forces and point of contact?
5. I am currently not using a Lagrangian-based solver, instead PenalityContactForceField. would changing to lagrangian multiplier be better? [ I realize that there are already discussions on how to get forces for this model and getting points of contacts should also be easy right?
   I really appreciate any help and discussion on this. Thank you!
   I have attached my code as well for reference

---

Environment

Context

• System: Ubuntu 20.04
• Version of SOFA: SOFA V24.06.00
• State: Installed from the website

Env vars

python -c "exec( \"import os, sys\nprint('#################')\nprint('--- sys.version ---')\nprint(sys.version)\nprint('--- PATH ---')\ntry:\n  print(os.environ['PATH'])\nexcept Exception:\n  pass\nprint('--- SOFA_ROOT ---')\ntry:\n  print(os.environ['SOFA_ROOT'])\nexcept Exception:\n  pass\nprint('--- PYTHONPATH ---')\ntry:\n  print(os.environ['PYTHONPATH'])\nexcept Exception:\n  pass\nprint('--- sys.path ---')\ntry:\n   print(str(sys.path))\nexcept Exception:\n   pass\nprint('#################')\" )"

--- sys.version ---
3.10.12 (main, Nov  6 2024, 20:22:13) [GCC 11.4.0]
--- PATH ---
/home/sun/sofa/build/bin:/home/sun/.local/bin:/usr/bin/python3:/usr/local/bin:/home/sun/sofa/build/bin:/home/sun/.local/bin:/usr/bin/python3:/usr/local/bin:/opt/ros/humble/bin:/home/sun/.local/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/snap/bin
--- SOFA_ROOT ---
/home/sun/SOFA/v24.06.00:/home/sun/SOFA/v24.06.00:
--- PYTHONPATH ---
/home/sun/SOFA/v24.06.00/plugins/SofaPython3/lib/python3/site-packages:/home/sun/SOFA/v24.06.00/plugins/SofaPython3/lib/python3/site-packages:/opt/ros/humble/lib/python3.10/site-packages:/opt/ros/humble/local/lib/python3.10/dist-packages
--- sys.path ---
['', '/home/sun/SOFA/v24.06.00/plugins/SofaPython3/lib/python3/site-packages', '/opt/ros/humble/lib/python3.10/site-packages', '/opt/ros/humble/local/lib/python3.10/dist-packages', '/usr/lib/python310.zip', '/usr/lib/python3.10', '/usr/lib/python3.10/lib-dynload', '/home/sun/.local/lib/python3.10/site-packages', '/home/sun/.local/lib/python3.10/site-packages/rerun_sdk', '/usr/local/lib/python3.10/dist-packages', '/usr/lib/python3/dist-packages']

[thundermms]

@hugtalbot & @bakpaul I am sorry for name-dropping like this, I know in the community you both are some of the most active people so I thought I could get your attention on this, I know I should be patient for the reply and not tag, so I sincerely apologize.

I am kinda stuck on the progress as I need to find the location of the point of contact between bodies to move forward onto my next part.
I have already implemented the Lagrangian solver and it seems to work fine, I just need to find the force applied on each body and the point of contact when the animation starts, however, I am not able to find that. I expected the point of contacts to be of the tetrahedral meshes of the collision body/visual body however I am not sure how to get it. Is there any reference code that I can get please

[bakpaul]

Hello !

Don't worry, a github notification is harmless... I'll answer in order :

1. It is in the worlds system of coordinate so yes it is cartesian. The origin is the same as the one of your models when loaded.
2. Those are the colliding elements ids. What is colliding here are not your mechanical objects but the surfaces represented by CollisionModels. Then the element type depends on the type of collision model you used. If it was LineCollisionModelthen this is the id of the edge in your collision topology
3. Well the difference is in the name. The element type defines what primitive is used when projecting one object on the other one. Ad you might think, the distance between two points which are 1D objects is always non zero. Which is no more the case between a Line and a Triangle for instance. Event using line v.s. line is better than using PointCollisionModel: the space representing a certain distance to the element is bigger and if you surface is closed then this space is contiguous. In summary havin PointCollsionModel is error prone, because interpenetration is easy to occur. Line is better, but holes could still appear when colliding against a PointCollisionModel. Line v.s. line is better because the resulting forbidden space is connected space. If you use surface, then it is the best because you'll have no hole in your surface, but it might lead to heavy projection computation.
4. & 5. Doing this you'll be able to find the ids of the points of your model subject to contact forces. Indeed using the element Ids, you'll manage to find the ids of the point (if it is an edge you can consider the two points of the edge as subject to contact force). You can do that using the edges data in your topology. But here because you use penalty-based contact, I don't know if there is a way to get only the force applied by the contact (@hugtalbot tell me if you think of anything). The only think you can do is to then take a look at the foce vector in the MechanicalState of you simulated node. Using the points ids as I explained, you can now see what force is applied on your object at this point. If no other force is applied to it, then you should get what you want. But in the end it'll indeed be easier to work with Lagrangian-based constraints

[thundermms]

@bakpaul Thank you very much for your reply, I understand the concept behhind now, but I am finding it difficult to implement it/get the expected outcome on sofa. I have also attached the code below for reference.

1. I tried using Trianglecollisionmodel for two models that I know have Trianglecollisions which is the skin and fat, and I tried to manually touch one on another, but I couldn't see any collision points ( the print statement was empty and it says No contacts detected!) I don't know why this is happening. I actually want to find the elementID's shown in the dof (mechanicalobject of both skin and fat) is there any way to get that instead of this.
2. I tried to use point and line between two models (rigid3d and vec3d )and when the interaction was being made I could see these items in the graph, is the elementID in this mechanical object is what was being printed when I use point and line in my initial question ?
   
3. In the code, the interaction between both is BilateralInteractionConstraint, is this correct or should something else be there ? could you also help explain how I should decide what to use for this particular case please
4. I have implemented lagrangian solver, in the mechanical object there are some forces for each elementID are these forces ( in newton) the correct forces applied on each element at any given time, can I use them or should I need to make any more difference? You mention it might be easier to work with Lagrangian-based constraints for force, can you redirect me to any documentation on this.
5. If i want to move a new third object of rigid3d to the normal of one of the elementId's of the mechanicalobject of the skin, how can i achieve this ?
6. Overall would you recommend any other improvements in the code that is being shared for the use case of simulation of skin and fat under a physical force
   github_discussion.zip