Error with Inverse Kinematics

[Brandon Brown]

Hello all,

I am using an upper body model to simulate a simple throwing motion. After running the scaling and kinematics, I get an error of around .09 meters in the marker which produces the maximum error during the motion itself. Prior to the motion, in the period of time when the subject is in the initial position, that error is less than the 4 cm which opensim says should be achieved to get good results. Does anyone know what the issue might be and also, why would the error go up during the motion itself? I have tried a few things including making sure that the coordinates are constrained to operate within their prescribed range values. I've also tried locking the thorax and turning off markers which I thought were unnecessary to the simulation. All of this did improve my results.


A question related to this, if I can't get the results improved to be within the 2-4 cm threshold during the movement, what effect would that have on computed muscle control and forward dynamics?

Thanks for any help anyone could provide.

Brandon Brown

[Ajay Seth]

4cm of error in the static case is excessive. 9cm during motion could be that the marker data and the model are not correctly expressed in the same reference frame or the model is a poor fit for the task. Is your model physically capable of reaching the marker in question? For example, if your model is scaled incorrectly, then it may not be able to extend far enough to match a marker on the hand. Or, if your model is constrained to planar motion then the error may reflect the fact there there is significant out of plane motion in the measured task. Have you looked at your model and the marker of concern at the time frame when the peak error occurred?