[QUESTION] Shape Model Scaling and Physics Simulation Units

[AmineElhafsi]

Hello,

I'm trying to use Warp in my pipeline to plan robot actions through physical interactions with a reconstruction of its environment. However, I'm finding that the size of some of my objects can be quite small, which I've observed can lead to instabilities in the simulation (e.g., small spheres will bounce off the ground plane at extremely high velocity at the start of the sim, meshes will rotate or bounce on the ground, etc.). Scaling up the model sizes has been the one solution to this that seemed to work for me. That said, I am trying to accurately simulate a real world scene and so this scaling will require me to adjust the units of other parameters in the simulation such as gravity, density, the material stiffnesses, and so on.

My first question is would doing so be sufficient for me to maintain coherence in the internal physics computations, or are there any subtle assumptions made internally regarding units? I understand that the simulation should be unit agnostic, however the default units appear to be kilograms, meters, seconds (given that the default gravity is -9.80665) so I'm wondering if there's anything I might miss in doing so.

For my second question, is there a global scale configuration parameter that I can set to automatically handle this for me? As far as I could tell there didn't seem to be, but I just wanted to check since this would make life a lot easier.

Thanksz1

[eric-heiden]

Hi @AmineElhafsi,

The simulation is indeed unit agnostic, we added sensible defaults for gravity, density and other material properties in the ModelBuilder class to make it easier to get started. These defaults are based on kilograms, meters, seconds. You are free to modify these values to use your own units; a global scaling parameter is currently not implemented.