I am trying to run the anybody with collected marker-based mocap data, and only found out that the parameter identification did not run normally. As in picture you see, the feet of the subject twisted in an abnormal angle. The ankle, or subtalar joint inversed nearly 90 degree.
The parameter identification can finish anyway somehow, but when running inverse dynamic, the programm failed immediately.
Do you know what is the possible reason, and how to fix that? Thanks a lot!
It is understandable that the inverse dynamics fails when you have bad kinematics.
From your image, it looks like you have only two markers on the foot: one at the heel, and one at the toes, isn't it? If this is the case, there is an indeterminacy in the model as the foot can freely rotate about an axis through these markers.
I can suggest adding some joint limits to the subtalar joint. This will prevent the extreme angles, but you may still end up with the foot at the actual joint limit.
Then, you can try adding weak mannequin drivers to the subtalar joint, which will try to keep the subtalar joint at 0 degrees.
Please add these drivers in the ExtraDrivers.any file that you can access through LabSpecificData.any
This might help you resolve the problem, but please be aware of the limitations, that is, it might not be representative of the actual movement, which we have no way of knowing.
