ID and Toy Landing model : huge moments and forces

[Joana Silva]

Hi everyone,

I collected data with Vicon and converted it with MotoNSM software, the .mot output has only the ground forces and torques(no CoP or PWA, is this wrong?)
I am having huge moments and forces when performing ID. here is part of the results :

IDresults.JPG (241.24 KiB) Viewed 2629 times

When previewing experimental data and sync. the motions everything seems ok... what do you think? : https://we.tl/oik5tDKssc

and the IK results with the adapted ToyLanding model ( I used this model because in the end I intend to compare the experimental GRF with the contact forces from Forward Dynamics on both calcn):
https://we.tl/eij6ZldT7A

Do you think that maybe the GRF are not well positioned?
- I applied to each calcn, defined as point force and expressed into ground. should it be between platform and calcn?
-I tried to disable the bushing forces and other forces (Lumbar extensionLimit and Knee Limits) to see if the forces decreased and it didnt work
- I ran RRA and the model falls down, i increased kp to 1600.0 and kv 80.0 and the model goes down and then up and then down again:
https://we.tl/0g4e2Y1wyq
- should I decrease the Ik, ID and RRA time interval for only the range when the subject steps on the force plates and the virtual platform?

I am running out of ideas to solve this problem, and where the problem might be...
Please let me know if you have something that can help me

Thank you!

[Joana Silva]

Also, my .trc and .mot files look like this, I compared with reference ones from available models to check the units, but I am still not sure whether or not the units are correct. Using MSL viewer the Forces are in N and Moments in Nmm, but I am still not sure about the header and units

trc.JPG (588.83 KiB) Viewed 2620 times

mot.JPG (469.43 KiB) Viewed 2620 times

thank you again,
Joana

[Erica Beaucage-Gauvreau]

I collected data with Vicon and converted it with MotoNSM software, the .mot output has only the ground forces and torques(no CoP or PWA, is this wrong?)]

The ground_force_px, ground_force_py, ground_force_pz should correspond to the COP where the force is applied.

Do you think that maybe the GRF are not well positioned?
- I applied to each calcn, defined as point force and expressed into ground. should it be between platform and calcn?

I believe that is correct. You can visualize where the forces are acting in the GUI by choosing "Associate motion Data"

What kind of motion are you looking at? Walking or Jumping?

[Joana Silva]

Hi Erica,

Thank you so much for the help !
-When I associate the motion data, the GRF vectors dont seem to be associated with any marker, like they are a bit far from the calcaneous, but synchronized in time.

-About the ID results, I am having forces at pelvis more than 10 000 N, that cant be right ...?!

-When I perform Forward simulation without actuator or external loads, I get a zero GRF in the midle and huge contact forces in heel strike and toe off:

forward.PNG (73.18 KiB) Viewed 2595 times

Maybe the problem is not having a contact sphere in the middle of the foot ? (only calcn and toes)

Do you think that using 2354 model and add the contact forces from Toy Landing model plus one more contact sphere would solve the problem?...

It is a walking simulation

Thank you so much

[Erica Beaucage-Gauvreau]

-When I associate the motion data, the GRF vectors dont seem to be associated with any marker, like they are a bit far from the calcaneous, but synchronized in time.

Did you load a motion first (results from IK) and then associate motion data? The green arrow will appear where the force is applied. See image here:

Force arrow.png (219.47 KiB) Viewed 2573 times

like they are a bit far from the calcaneous

That might be why forces are big? It should be aligned pretty well I believe.

[Joana Silva]

Thank you for you help Erica
I have done that previously but only for the dynamic trial,where the green arrow seem to be a bit dislocated but following the walking properly:

dynamic association.PNG (107.25 KiB) Viewed 2557 times

But in fact doing for the static pose is totaly wrong ...Do you know how can I fix this?

static association.PNG (71.08 KiB) Viewed 2557 times

Do you believe the axis are wrong?

axis.PNG (70.94 KiB) Viewed 2557 times

When using MOtoNSM I used this Laboratories protocol :

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<Laboratory xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<Name>VT</Name>
<MotionCaptureSystem>Vicon</MotionCaptureSystem>
<CoordinateSystemOrientation>YZX</CoordinateSystemOrientation>
<NumberOfForcePlatforms>2</NumberOfForcePlatforms>
<ForcePlatformsList>
<ForcePlatform>
<ID>1</ID>
<Brand>Kistler</Brand>
<Type>2</Type>
<FrameRate>200</FrameRate>
<FPtoGlobalRotations>
<Rot>
<Axis>X</Axis>
<Degrees>180</Degrees>
</Rot>
</FPtoGlobalRotations>
</ForcePlatform>
<ForcePlatform>
<ID>2</ID>
<Brand>Kistler</Brand>
<Type>2</Type>
<FrameRate>200</FrameRate>
<FPtoGlobalRotations>
<Rot>
<Axis>Z</Axis>
<Degrees>90</Degrees>
</Rot>
<Rot>
<Axis>X</Axis>
<Degrees>180</Degrees>
</Rot>
</FPtoGlobalRotations>
</ForcePlatform>
</ForcePlatformsList>
</Laboratory>

And the lab looked like this, with Z pointing up and Y forward :
Static trial :

staticMokka.PNG (6.88 KiB) Viewed 2557 times

Dynamic trial:

dynamicMokka.PNG (5.35 KiB) Viewed 2557 times

[Joana Silva]

Hi again,

I have figured it out already the conversion from the FP referencial to the lab was not correct, guess I had just follow the example and didn't change it for my FP!
Now it looks good :

working.PNG (64.62 KiB) Viewed 2540 times

I also filtered the marker trajectories and forces


Thank you so much

[Erica Beaucage-Gauvreau]

Great! Good job. Did it solve your high forces at the pelvis?

[Joana Silva]

Hi Erica,

No it didn't, they are still ridiculous

pelvis.PNG (37.17 KiB) Viewed 2509 times

pelvissum.PNG (26.77 KiB) Viewed 2509 times

And when performing RRA I get strange recommended adjustments, like almost zero and negative...
************************************************************
* Recommended mass adjustments:
* Total mass change: -78.5788
* ground: orig mass = 0, new mass = 0
* platform: orig mass = 0.100313, new mass = -0.00409063
* pelvis: orig mass = 11.8138, new mass = -0.481753
* femur_r: orig mass = 9.33049, new mass = -0.380486
* tibia_r: orig mass = 3.7191, new mass = -0.15166
* talus_r: orig mass = 0.100313, new mass = -0.00409063
* calcn_r: orig mass = 1.25391, new mass = -0.0511329
* toes_r: orig mass = 0.217277, new mass = -0.0088603
* femur_l: orig mass = 9.33049, new mass = -0.380486
* tibia_l: orig mass = 3.7191, new mass = -0.15166
* talus_l: orig mass = 0.100313, new mass = -0.00409063
* calcn_l: orig mass = 1.25391, new mass = -0.0511329
* toes_l: orig mass = 0.217277, new mass = -0.0088603
* torso: orig mass = 34.3437, new mass = -1.40049

Do you think it has something to do with the properties of the HuntCrossleyForces ? How do I know if it is related with units incompatibility?

Thank you again

[Joana Silva]

Hello again,

The huge values were due to penetration between the foot and platform:
this post helped me : viewtopicPhpbb.php?f=91&t=7784&p=0&start=0&view=
And despite visually is not touching the platform, I believe the forces are being exerted, the question it would be how to avoid the penetration not just by moving the platform down and validate the research in the future ...

Also, I followed Scaling-->IK-->ID-->SO-->Forward dynamics, is that ok using SO instead of CMC ?

Thank you !