Page 1 of 1

Adding DOF to translation

Posted: Thu Jul 31, 2014 5:25 pm
by liltak
Hi,

First I would like to appreciate the team for creating the model.
I have been looking for this kind of model that I can use.
I loved this model for our research and I would definitely love to use it.

But unfortunately I have a problem when scaling the model to the motion capture markers of our data on OpenSim. It seems that the problem defined in the link below is happening to the model.
https://simtk.org/forums/viewtopic.php? ... 7666#p7666
The problem shown on the link induced that these problem happens when the model was not able to translate.
Since I would like to use this model very much, I would like to know how to add the DOF to translation to this model.

Best,

Re: Adding DOF to translation

Posted: Fri Aug 01, 2014 6:55 am
by bogert
Dear Takayo,

We have not used the model for inverse kinematics, but I think I understand the problem.

In our model, the thorax is attached to the ground and can't move. In your subject data, the thorax is probably positioned differently, but the model can't match that position because the thorax can't go there.

The solution would be to add 6 degrees of freedom (3D translation and rotation) to allow the thorax to move. Replace the weldjoint between thorax and ground with a connection that has 6 degrees of freedom. If you need an example of such a connection, load one of Opensim's gait models and see how they did the connection between pelvis and ground with 6 DOF.

If you're thinking of doing dynamic analysis, you should not use the residual reduction algorithm (RRA). RRA considers all forces between thorax and ground as residuals that are supposed to be reduced to zero. This would not be correct for this model.

I am not an experienced Opensim user, hopefully this gives you enough information.

Ton van den Bogert

Re: Adding DOF to translation

Posted: Fri Aug 01, 2014 4:23 pm
by liltak
Dear Ton,

Thank you for your reply it seems I made it.

Also thank you for the advise for analysis.
Since I am very new to this kinds of research, I have questions about your comment on the analysis too.
Does that mean this model does not have the leg to support the body, which makes it impossible to analyze the motion of the arm itself using RRA or Static Optimization?
Do you think that problem will be solved if I cancelled the movement induced by the movement of the thorax by using the other motion capture marker's position?(e.g. position of the marker on the chest, which has less effect of the arm motion)

I am sorry for a amateur questions...

Best,
Takuya

Re: Adding DOF to translation

Posted: Sun Aug 17, 2014 3:18 pm
by bogert
RRA is not appropriate for this model. It is usually used for gait where there are several degrees of freedom without actuation.

Static optimization also has some issues with this model, because Opensim is not aware of the forces between scapula and thorax. We are working on Matlab code to perform muscle force estimation that takes everything into account.

Subtracting the thorax motion from all markers is not a bad idea. You would need at least three markers on the thorax. Use those to define a coordinate system, and then transform all marker trajectories to that coordinate system. Then finally you need to transform to the thorax coordinate system used in the model. I am not sure how that was defined (other project members may know) but you can see it in Opensim if you turn axes on.

Ton van den Bogert