ID: same pelvis residuals with/without GRF

[Joana Silva]

Hi everyone,

My goal is previewing GRF and compare them with the experimental ones.
I am having huge moments and forces when I evaluate my simulation through ID, so I performed ID without GRF to see the residuals at pelvis:

When I simply use 2354 model :

pelvis residuals with no contact forces.png (19.83 KiB) Viewed 955 times

but when I use a model with contact geometries on foot and platform

contactmodel.png (19.29 KiB) Viewed 955 times

it doesnt make difference and I get almost the same with and without applying GRF...

I really have no clue why this is happening, what's wrong?

[Thomas Uchida]

it doesnt make difference and I get almost the same with and without applying GRF...

If I understand correctly, you're computing ground reaction forces using a contact model and also applying ground reaction forces that were measured experimentally. These modeling choices are mutually exclusive, so the reported behavior doesn't seem unreasonable.

[Joana Silva]

Thank you Tom,

So you are saying that if I want to preview the contact forces with (Forward Dynamics) I should never at any step give the GRF as inputs?!

When I run RRA without external loads and adjusted the model it goes like this: https://we.tl/Jm1XcXeMxy

but if I run CMC keeps on giving me errors like :
(...)for many muscles:
CMC::Warning: CMC cannot compute controls for muscles with muscle controls < 0.02.
The minimum control limit for muscle 'extobl_l' has been reset to 0.02.
WARN- a desired points file was not specified.

CMC.computeControls: t = 1.75
SimTK Exception thrown at InteriorPointOptimizer.cpp:261:
Optimizer failed: Ipopt: Infeasible problem detected (status 2)
OPTIMIZATION FAILED...

CMC.computeControls: ERROR- Optimizer could not find a solution.
Unable to find a feasible solution at time = 1.75.
Model cannot generate the forces necessary to achieve the target acceleration.
Possible issues: 1. not all model degrees-of-freedom are actuated,
2. there are tracking tasks for locked coordinates, and/or
3. there are unnecessary control constraints on reserve/residual actuators.

is because I have a platform? because once locked shouldn't interfere right? I tried everything I remembered. How can I solve the part "CMC cannot compute controls for muscles with muscle controls < 0.02." ??

I also checked the ID results after RRA and the moments and forces are still really huge :

IDafterRRA_noGRF.PNG.png (11.64 KiB) Viewed 921 times

Would you help me to figure it out what's wrong?
Thank you

[Thomas Uchida]

So you are saying that if I want to preview the contact forces with (Forward Dynamics) I should never at any step give the GRF as inputs?!

No, based on your initial post you "use a model with contact geometries on foot and platform" and "get almost the same with and without applying GRF". It doesn't make sense to apply experimental GRFs to a model that already has a contact model for computing the GRFs.

There are some tips at the bottom of the "Getting Started with RRA" page in the Confluence documentation (http://simtk-confluence.stanford.edu:80 ... d+with+RRA). See, for example, item 3 in the "Troubleshooting" section:

"If residuals are very large (typically, this is greater than 2-3 times body weight, depending on the motion), there is probably something wrong with either (i) the scaled model, (ii) the IK solution, or (iii) the applied GRFs. To double-check that forces are being applied properly, visualize GRFs with IK data (you can use the Previewing Motion Capture (Mocap) Data function in the GUI)."

[Joana Silva]

Thank you so much for the help Tom,

I understand that the contact geometries will then replace the need for GRF as input

And about the previewing of GRF they seem not to be associate to any foot marker but they follow properly the walking motion. But when I did the association between the static pose and the MotoNMS elaborated GRF (with Data processing and not StaticElaboration, can I even do this?! because the StaticElaboration creates only .xml and .trc files ) and it is dislocated in fact :
viewtopicPhpbb.php?f=91&t=7779&p=20699&start=0&view=

But the markers are well aligned, do you have a guess about the problem?

Sorry for insistence but it is the first time I am dealing with all this software and details and I really don't know what is going on

[Thomas Uchida]

Sorry, I'm unfamiliar with MOtoNMS. Perhaps someone else on the forum can help. You could also check the MOtoNMS forum (indexPhpbb.php?group_id=927). See, for example, this topic: viewtopicPhpbb.php?f=927&t=6918. Also note that there are working examples on Confluence (http://simtk-confluence.stanford.edu:80 ... +Tutorials) and in the OpenSim Models directory.