ID: ERROR - mass matrix is singular

[Marc Carmichael]

Hello,

I am performing an inverse dynamics simulation with the Stanford upperlimb model. When I load my .mot file I get the error "InverseDynamics: ERROR - mass matrix is singular".

My ideas for the possible cause of the error are the following:

1 - I made an artificial mot file (artificial in the sense that it was not produced by inverse kinematics using real marker data) which smoothly moves the arm into different poses, with little pauses at each pose. Perhaps when at each pose, since the coordinates do not move at all, could this be interpreted as an error??

2 - I have only specified mass and inertia properties for some bodies, not all of them. If the inverse dynamics tries to then use an empty mass/inertia matrix, would this cause issues??


Or perhaps there could be another issue I am not thinking of. In any case, can you please let me know which is likely to be the cause of the error, and what steps I can take to get around it?

Thank you

Marc

[Ayman Habib]

Marc,

The test for mass matrix does not take into account the specific motion. You should have mass properties specified for all bodies in your model (except for those rigidly attached to bodies with mass and ground). This involves specifying a value for the mass in the .osim file and also specifying non-zero Inertia values (Ixx, yy, zz).

Hope this helps,
-Ayman

[Marc Carmichael]

Thanks, I wrote a simple program that cycles though each body and gives it an infinitesimally small mass and inertia matrix. After this the ID worked.

Cheers

[Nur Adila Faruk Senan]

Hi,
I'm getting the same error message as Chris, only all my bodies (excluding ground) have non-zero masses and moments of inertia.

This only happens with the model of the lumbar spine that has 3 DOFs but 12 constrained coordinates (i.e. model Constrained) and not the model that has 15 DOFs (with no constrained coordinates) and bushing forces to mimic the intervertebral disc stiffness (i.e. model UnConstrained)--inverse dynamics works fine with the latter but keeps throwin up the "Mass Matrix is Singular" error message with the former.

I'm not sure if it's because the model has muscles or if its some other issue because I previously had another model that was Constrained, but didn't have any muscles, and that worked fine --> the output was a .sto file with the flex_ext, lat_bending and axial_rot moments (i.e. 4 columns, with the first being time) although that model had the same coordinates as my current Constrained model.

Thanks for the help,
adila

[Nur Adila Faruk Senan]

just figured out that i have wrapping surfaces in there that were massless - made them non-zero (but tiny) and that seems to have fixed the problem.

[Ayman Habib]

Adila,

Wrap Objects do not have mass, it must be something else that you fixed. It may suggest that the first mass you specified was not described correctly in XML. If you continue to have problem with this please let me know.

-Ayman