Parameter optimization of dancer

[Bharathhs (Bharath)]

Hello again,
I have an issue after running parameter optimization, my thigh segments become longer and the shank segments become smaller and the final model has longer thigh bones and shorter shank bones. Since i dont have markers on head and trunk region. The optimization is set to off in labspecificdata.any. In order to improve the optimization process near the thigh regions, i also locked the parameters in the upward direction preventing it from optimizing. so i could have the length remain the same and not get longer. It still ends up with long and slender thigh bones which is not the case with my real time dancer(c3d file input).
-I would like to know if the marker placements actually place a vital role in my issue, would moving the marker placements further up near the thigh region improve the optimization and prevent long thigh bones.
-and also how should we understand that a certain marker need not be optimized in certain direction. I do understand that the marker placements are initial guess to the optimization. I would like to know to what extent i should get the initial guess of the markers right !.
-Also the markers in the upper body plays a important role to the optimization, is there a standard set of markers placed on the body that needs to be optimized, i mean the direction of a specific marker. this would be interesting.

Before running parameter optimization

After completing paramter optimization

inverse dynamics

my AnyBody file :
https://liuonline-my.sharepoint.com/:f:/g/personal/bhahu521_student_liu_se/Erf9XUvYx9ZJmFHPtcH-MfkB2sjtJm7FeX_UO92z_DIwow?e=6nDAl7

the directory is AMMR-applications-Mocap examples-Plugingait simple- full body main.
the c3d file can be found in the input file directory.

your help would mean a lot to me.

[toerholm (Søren Tørholm)]

Hi Bharathhs,

When looking at the picture before the optimization, it can be seen that the thigh markers are place too high compared to the recorded positions. Secondly the height position of the thigh markers are locked they can not adjust up/down.

I think there could be two ways to solve the problem

1. Free up the height direction to be optimized. there is a risk that the optimization problem will become undetermined, but it might just work
2. Alternatively adjust the maker position to have a more realistic height position and leave the height variable locked..

Best regards
Søren

[Bharathhs (Bharath)]

hello

1. the reason i made this change was because the optimization gave me the same result with the markers on the thigh free .
2. so now i have to adjust the marker position according to the c3d marker position or the real time marker position. when i meant the real time marker position its from the video footages during recording motion.
   
   
   
   
   according to the 2 pointer from you, you say i have to keep the markers according to the c3d data input, as now it place a bit higher. but according to the reality (actual footage) it is place a bit higher.

[toerholm (Søren Tørholm)]

Hi Bharathhs,

I see the problem,

I have tried to run a bit with the model and tried to lock directions in the clusters, but it was not enough to prevent the thigh for getting too long.

So i have added this constraint in the labspecific.any file.

Main.Studies.ParameterIdentification ={
  
  //This is a crude estimation...
   AnyDesMeasure ThighShankLength_Equal = {
      Val = 1.0*.RightThighLength.Val - 1.0*.RightShankLength.Val;
      Type = EqualToZero;
    };

};

This is a constraint which will make the shank and thigh equal in length, you can change this line to a different ration if you like.

The problem is that you are lacking a knee marker, it is possible that a if you had a trial with a lot of knee flexion it would have worker without this extra constraint.

The constraint above added to the labspecific.any file seems to work ok.

Best regards
Søren