simtk.org

Apologies for the quite simple question but I seem to have a gap in my understanding somewhere that I can't figure out. Basically, I want to show that the summed contributions of muscle forces & reserve actuators matches the inverse dynamics solution for a set of motion states obtained from CMC.

Inputs: states & controls from a CMC simulation, CMC actuators XML file, external forces XML file.

Steps:

• Compute inverse dynamics from CMC states with external forces.
• Run Muscle Analysis on CMC states & controls, specify moment arms to be computed, same actuators & external forces as CMC.
• Run Force Analysis on CMC states & controls, same actuators & external forces as CMC.
• Compute the generalised force at each joint by summing the contributions from each muscle (force * moment arm), as well as contributions from reserve actuators.

I would expect the generalised forces from both sources (inverse dynamics, muscle forces) to be identical. However, I am getting slight differences in my results.

See results here: https://ibb.co/ZGvJkTn

Can anyone tell me if I'm making a mistake in the above pipeline or forgetting to account for some forces somewhere?

Kind regards,

Daniel

Hi Daniel,

The steps are:

1. Perform inverse dynamics analysis.
2. Execute CMC. I think it is possible to append a MuscleAnalysis in the .xml setup file of CMC so you will not have to run the AnalyzeTool separately. Make sure that the MuscleAnalysis calculates the joint moments and not the moment arm. In this way, you ensure that you have not misconfigured the analyze tool.
3. Compare the moments calculated by the muscle analysis with the one computed from inverse dynamics.

Note that you might not get a perfect match, because CMC may not track the kinematics perfectly. Please remember that CMC is a tracking method.

Hi Dimitar,

Thanks for the help. However, we are running inverse dynamics on the CMC_Kinematics_q file that is output from the CMC simulation - in this case, wouldn't we expect the ID torques to be equivalent to the sum of the muscle force/reserve actuator contributions, since it is computed for the motion produced by CMC rather than the original motion?

Daniel, your logic sounds correct to me (although I'm not a CMC expert).

One possibility is that maybe the "muscle forces" you are using are the "active fiber forces" or "total fiber forces"? For comparison with inverse dynamics moments I think the "tendon forces" should be used.

Another possibility: maybe the "residual forces" (different thing from the "reserve actuator" forces) are generating moments at the joints that are not accounted for in the analysis?

Ross

Hi,

Just to complement Ross's repose. I think the difference is probably caused by the reserve actuators used in CMC. Did you add their contribution to your moment from ID?

Also, if you change the kinematics in ID then maybe you have to update the .xml file of the GRF which also use the kinematics. However, I would not use the kinematics from CMC in ID. The purpose of this approach is to compare the error in the moments due tracking and reserve actuators. It cannot be exactly the same.

Furthermore, based on your initial post, you do not have to calculate the joint moments from the product R * fm manually. This is done by the MuscleAnalysis if you enable the calculate moment option.