Optimization of Pelvis width, foot length

[v22 (Vivek)]

I had a question regarding the optimization of the width of the pelvis and the foot length in the full body model. While running a few simulations, I observed that some of the markers seem off the foot, onto the side as shown in Fig 1. I am inclined to think that this is due to the fact that the width of the pelvis is the same for all subjects in the simulation, which may not necessarily be the case. Similarly for the length of the foot, which may cause the reaction forces at the ankle to change by quite a bit. Should these be turned on as this length will vary from subject to subject? I was wondering if anyone has an idea of how much the results can be effected due to this.

[Bjorn (Bjørn Keller Engelund)]

Hi @v22

It is recommended to keep all posts regarding the same model /issue in one tread.
That makes it much easier follow for us and other users.

I agree that the showed marker seems out of place.
But it seems that position is off in the data file (indicated by the red marker)

Have you checked the data file that it is placed as expected ?
Have you moved the blue marker to accommodate the position that you placed the red marker?
What directions are you optimizing in the ParameterIdentificationStudy?

Regarding the pelvis width, are you not scaling that based on your marker input?
Have you removed the pelvis width from the scaling procedure?

Best regards,
Bjørn
AnyBody Technology?

[v22 (Vivek)]

Hello Bjorn,

I apologize for the cross posting.

I came across something after speaking with a consultant from AnyBody. One of the possible issues could be that in the experiment, we have the subject running with shoes on, which is different than the settings in the AnyBody model, where the person runs bare foot.

Have you checked the data file that it is placed as expected ?
Yes, this seems to be correct. But this marker is placed on the shoe.

Have you moved the blue marker to accommodate the position that you placed the red marker?
Yes I did. After moving the markers close, I ran the motion parametrization study.

What directions are you optimizing in the ParameterIdentificationStudy?
For the marker that looks out of place, I am not optimizing in any direction(part of the reason for doing this is that the marker was placed on the shoe).

Regarding the pelvis width, are you not scaling that based on your marker input?
We have that setting turned off in the model, which I think should be ON.

Have you removed the pelvis width from the scaling procedure?
That is correct. However, I think that the width of pelvis and length of foot must be allowed to scale based on the markers.

With regards to the other questions, here is what we are currently doing.

What tolerance are you altering?
In the Kinematics.any file;

InitialConditions.KinematicTol=1e-3; to 1e-6
Kinematics.KinematicTol=1e-3; to 1e-6
ParameterOptimization.ConvergenceTol=1e-2; to 1e-6

In the InverseDynamics.any file,
ParameterOptimization.ConvergenceTol=1e-3; to 1e-6

My understanding was that the change in SrelOpt will move the red markers(the one that come from the AnyBody model) and not the blue ones. So I was not sure if we can move them around, since they come from the experiment. Is changing SrelOpt the way to do that?

Thanks,
Vivek.