I tried to calculate JRFs at the hip/knee/ankle during running using a subject-specific C3D data. The subject has a mass of 67kg, and height of 1.68m. She runs at 3.6 m/s on a treadmill.
Despite that no problem was encountered wrt the simulations, the reported JRFs from IDF seem to be too low. As seen in the attached pictures, the leg forces are between -2000 to 3000N (3-4.5 BW). To me the values seem to be too low - more like the values i would expect during walking conditions. I also observed low muscle activities in the lower extremity in the model.
I was trying to play with the StrengthParaeters (SpecificMuscleTensionSpine, SpecificMuscleTensionShoulderArm, StrengthIndexLeg) but didn’t get very far.
Just would like to add a few more notes if it helps:
After checking the residue force as suggested on wiki (http://wiki.anyscript.org/index.php/All_about_Kinetics), the force between the trunk and the global environment reached 1.5E3 N. It seems that the model artificially added this amount of force to balance things out, leading low forces in legs. If that’s the problem, how do I fix it?
I would check these things:
[ul]
[li] Load on feet’s are these applied correctly? Try to run model with no muscles and check the joint reaction forces, if the load is being applied
[/li][li] Direction of gravity is it correct?
[/li][/ul]
As you suggested, the load wasn’t applied correctly on the feet. The problem seems to stem from the automatic detection in ForcePlateType2. After changing it I got smaller residue force (upper figure) and more reasonable JRFs (lower figure). However, the residue force still remains -200~300N. The peaks of the residue forces coincided with the time when the foot hit the FP.
I checked the mocaprunner model, whose residue force was also over larger than 200N.
200 N is still a bit high side and could come from
[ul]
[li]Missing arms in the analysis
[/li][li]Mass properties differences between model and subject
[/li][li]Measurement errors.
[/li][/ul]
the more dynamic the trial is there more these effects will impact the model.
If the error is largest at heel strike it can be caused by inaccurate accelerations of the body which needs to be balanced out by the measured forces… So filtering of both motion and forces will impact the residual you will see.
Since I did a full-body running simulation, missing limbs shouldn’t be a causing problem. I used AnyWomanExtPercentile Scaling. However, I changed the definition of AnyVar Body_Mass in AnthroData so the model has the correct body height and weight. as the woman subject The change is not substantial, though (i.e., 1.2 kg).
I am more inclined to believe the error comes from the dynamic events such as deceleration and acceleration during running. The largest component (top panel in the attached figure) is PelvisPosZ (proximal-distal direction), followed by PelvisPosX (medial-lateral direction). I tried to change the Fc and CutOffFrequency in C3DSettings (N=2, Type=LowPass) but couldn’t get the residue force under 200N.
Another question: when shall we make changes to StrengthParaters? I used the default values (90 for SpecificMuscleTensionSpine and 1 for StrengthIndexLeg) and not sure if the values are proper for my case (healthy female).
Concerning the strength scaling i would look at the overall activity level and then possible down scale the max stress number accordingly. This number can be used to obtain the individual strength of the subject. The best would always be to do an individual strength measurement of the subject and adjust the strength accordingly.
Thank you for your reply! Do you mean changing AnyVar MaxStress? Isn’t it only for muscle coloring? Can you provide a little more details on how to improve the values of muscle strength?
I checked the StrengthScale of Shank in my model, which is 0.52; while that in the MoCapRunner model is 0.84. I understand it’s difficult to quantitatively specify muscle strength for individuals without experiments - can you provide an estimate for “strong” vs. “week” muscles? i.e., is a value of 0.52 considered average strength for a subject with such body composition?
thank you!
p.s. regardless how hard I tried, I still cannot keep the residue force under 100. In the attached picture, you can see 6 peaks, corresponding to 3 running cycles.
The strength scale values are calculated automatically based on the weight and height of the person, these can not really be altered without changing the scaling law which is also an option.
The strength can be altered in the humanmodel.any there is ususlaly a section like this:
by adjusting these values you can make the model stronger or weaker in body sections.
Concerning the residual …
Do you have any trials that are less dynamic… possibly a static trial … this would allow you check that you had the force plates are reading out correct weights…
A simple test like placing a static weight on the plate and read out the values would allow you to see if forces are measured and applied correctly in the model.
Have you tried to visualize the CoP and the direction of the GRF acting from CoP did it look ok?.
By increasing StrengthParameters (e.g., SpecificMuscleTensionSpine or SpecificMuscleTensionShoulderArm), the calculated muscle activities (Muscle Force/Muscle strength) will be lower, is that correct? By changing StrengthParameters, predicted muscle forces & JRFs remain the same, is that correct? Moreover, if StrengthIndexLeg=2, does it mean the strength of leg muscles are twice as the Spine muscles?
I still cannot keep the residual force under 200N. The GRFs measured from the static trial (left panel of the attached figure) show some fluctuations but not too severe. By visualizing CoP (right panel of the attached figure), I do see a small amount of GRFs on the FP where there should be no actual load applied. However, the magnitude of the force seems small.
[ol]
[li]For the static trials i can see the forces measured, but what is the residual for these models… also 200N? The residual for a static model should be significantly lower…since there are no dynamics impacting the model.
[/li]
[li]On the ModelView i can see a force acting in the corner of the plate is that the CoP ?.. please verify that when displaying the CoP it will be located within the foot.
[/li]
[li]Concerning the scaling if the index is 2 it doubles the strength of these muscles
[/li][/ol]
For the static trials i can see the forces measured, but what is the residual for these models… also 200N? The residual for a static model should be significantly lower…since there are no dynamics impacting the model.
The subject weighs 601N. The total of the GRFs measured by both FPs is between 620-645 N (see the attached picture). So the difference (residue) in the static model is between 20-45 N. Is this value still too high?
On the ModelView i can see a force acting in the corner of the plate is that the CoP ?.. please verify that when displaying the CoP it will be located within the foot.
I am puzzled by that CoP, too. At this time point, the CoP of the right foot shouldn’t show. Is this related to the errors in #1? (i.e., Cop positions/magnitude should be improved if the residue problem is addressed?)
Concerning the scaling if the index is 2 it doubles the strength of these muscles
So the activity will be half lower (the color of the muscles will be darker/bluer?)
1: This seems resonable, though it could indicate that the residuals could be improved a bit by altering the mass to be in closer to what the force plate measured.
2: The CoP needs to located within the foot when there is a load applied. When the there is no contact on a plate CoP becomes “undeterminate” and it will go to origo i think.
If you have access to the experimental setup maybe try a simple setup where you place a fixed weight in a well specified location on the plate and then try to see if this is represented in the model correctly.
