[Question] Question about deformable body material

[yuezhezhang]

Question

Hi,

I try to test the behaviors of deformable objects in file run_deformable_object.

As I searched online the poisson ratio and Young's module of steel should be 0.3 and 200GPa, but when I set poissons_ratio=0.3, youngs_modulus=2e11, it does not seem like steel:

run_deformable_obj.mp4

I also tried to pick the object, and the gripper also seemed to smash it.

Do you know how may I tune the behavior better? Thanks in advance.

[Mayankm96]

From what the PhysX team mentioned -- For higher stiffness, you can:

• lower the voxel mesh resolution (=faster convergence)
• lower the time step
• increase the iteration count
• maybe enable option "external forces per substep" but it does not help that much
• Use the TGS solver (but that's the default solver in kit anyway)

Also a soft body will always be a bit soft, no matter how hard you optimize the solver. For real rigid bodies, just use the PhysX SDK's rigid bodies.

[yuezhezhang]

Hi Mayank,

thanks for the feedback!

I searched the code and found relevant properties of soft body may lie in the functions modify_deformable_body_properties of source/extensions/omni.isaac.lab/omni/isaac/lab/sim/schemas/schemas.py and add_physx_deformable_body of _isaac_sim/extsPhysics/omni.physx/omni/physx/scripts/deformableUtils.py.

Specifically, in add_physx_deformable_body, I guess these arguments correspond to what you mentioned:

• simulation_hexahedral_resolution: voxel mesh resolution
• solver_position_iteration_count: iteration count

I guess I am correct since IsaacSim 4.2.0 lowered simulation_hexahedral_resolution from 10 to 1 and updated solver_position_iteration_count from None to 10. This change is in line with your suggestion.

However, sadly, it is still a bit soft though I tried to tune the poissons_ratio, youngs_modulus, dynamic_friction as this.

I may leave this discussion as it is now, if someone has new findings or discoveries, please hit me up!