Verifying IAA induced constraints

[Adam Yoder]

Hi all:

Attempting to do some verification of induced constraints (rolling-without-slipping) versus experimental GRF, during walking - forces look good, but having issues with moments

From the induced_constraint_reactions.sto output file I'm pulling all foot_ground entries for forces/moments (e.g total_right_foot_contact_ground_My)
(dotted lines on plot)
From my experimental input_grf.mot, I'm pulling forces, and the non-zero free moment My - which is in the +/-5Nm magnitude range
(solid lines on plot)

Since im applying rolling, I would expect Mx and Mz to come out as zero on each foot in the constraints if applied correctly, since the foot is free to rotate in those direction at the point of application?
And for the free moment My about the vertical axis, should fall near +/-5Nm order of magnitude
Although as you can see in below image Mx/Mz for CON are nonzero and at unrealistic magnitudes exceeding 1000Nm

Perhaps unit scale issue or mismatched reference frames? Osim output file notes units are SI N/m/s
Or the constraint is not being applied in the correct location?
Thoughts?

GRF_comparison.jpg (274.67 KiB) Viewed 1246 times

[Nicos Haralabidis]

Hi Adam,

Congratulations on the analysis so far! The constraint reaction forces appear to match the experimental reaction forces extremely well! Good job!

I haven't yet myself ran the IAA using the rolling constraint - that is my major objective for the week - but I have been spending some time familiarising myself with the analysis.

To my knowledge, as you also mentioned, the Mx and Mz moments should be zero as the rolling constraint doesn't permit Mx and Mz moments to be developed.

Have you compared your setup file with an example IAA setup file? Like the one in Sam Hamner's project? What values have you input for the a) rolling body, b) surface body, c) surface normal, d) friction coefficient and e) contact radius?

I can't imagine it's a problem between the reference frames, as you would have noticed this earlier in your analyses?

Have you made any further progress with troubleshooting this? I would be extremely interested in knowing!

Kind regards,

Nicos Haralabidis

[Ajay Seth]

Adam,
Remember that the constraint reaction forces are presented in the body frames (e.g. calcaneous and/or ground) and are not resolved at the point of contact. If your resolve the moments to the effective center-of-pressure (CoP) on the floor plane that will enable you to verify the CoP and validate the vertical reaction moment against your experimental values.
Cheers,
Ajay

[Adam Yoder]

Thanks Ajay - resolving moments from ground to COP using experimental px,py,pz solved a majority of the confusion!
New plots in normalized units below
Vertical My looks to be predicted well

Nicos: Looks like Mx & Mz constraint moments are still generated, to a small degree in later stance. I am not sure how this is possible, given the definition of rolling, as you pointed out - perhaps something to do with remaining, small offset between simulated vs. true foot & COP position after IK+RRA ? e.g. Shown in image is model at 100% stance at left heelstrike

GRFvsEXP2.PNG (98.38 KiB) Viewed 1160 times

LHS.PNG (127.41 KiB) Viewed 1160 times