Heavily rounded body parameters in bsm.osim

[auberger]

Hi Marilyn!

I'm currently implementing center of mass calculations to compare CoM linear and angular acceleration and with my own predictions of ground reaction forces and moments for validating dynamic consistency in human motion (Newton's laws). I just realized that segment mass, CoM center and inertia in bsm.osim are heavily rounded compared to tmp.osim (I assume that bsm is a scaled version of tmp). Comparing the two shows significant rounding:

• "lumbar_body": {"inertia": [1.0, 1.0, 1.0, 0, 0, 0]}
• "thorax": {"inertia": [1.0, 1.0, 1.0, 0, 0, 0]}
• "calcn_l" mass: 1.25
• Other segments show similar rounding patterns

Questions:

1. Is tmp.osim the generic model that was scaled to create bsm.osim and does the bsm model in models/skel_models_v1.1 correspond to beta=0 SKEL mesh?
2. Were the inertia values intentionally rounded to 1.0 for simplification?
3. Could you maybe provide the original (unrounded) version of bsm for more precise calculations or share with me how you scaled the model so that I can replicate it?
   Thanks in advance!

[MarilynKeller]

Hi!

1. No it is the opposite way. bsm.osim was fitted to a subject to get tmp.osim. tmp.osim was actually a temporary file created when testing the release, it should be ignored. bsm.osim results of stitching existing .osim body models togethers plus implementing a custom spine joint. The scaling and inertials parameters it has are inherited from these models and this do not correspond to SKEL(beta=0). To find that correspondance, one could fit SKEL of bsm.osim joint locations to find that beta. or fit bsm.osim to a sequence of markers extracted from a motion sequence of SMPL(beta=0).

2,3. This might be something done by Nimblephysics during the optimization of fitting bsm.osim to a subject's markers sequence. You can check the optimization code here. Note that AddBiomechanics optimization has evolved since I generated the fits, so maybe if you use it to run a fit of bsm.osim to a sequence now, the inertial parameters may look different.
You can try this here

Hi hope this helps!