OpenSim models vs 6DOF

[Mohammadreza Rezaie]

Dear experts,

I would like to know your opinions about the Opensim models and the 6DOF models when running IK.

For example in gait2392 model, the hip joint has only 3 degrees of freedom (all rotations). If I add 3 translations and create a 6DOF hip joint (as well as knee joint), the thigh segment sometimes goes beyond the acetabulum fossa and the knee joint space changes drastically. I attached a video to show this better. Please note the hip and knee joints and the segments distances.

https://drive.google.com/file/d/1D8VD9A ... sp=sharing

This is exactly the same as Visual3D 6DOF models.

Please let me know about the advantages and the disadvantages of this 6DOF models for solving the inverse problems in OpenSim compared to the conventional OpenSim models.

Cheers,
Mohammadreza

[Dimitar Stanev]

Hi Mohammadreza,

By unlocking the joint, you get better IK tracking. However, think of this as over fitting. With this approach you are capturing the noise and skin artifacts. It is best to keep the model constrained even if it is not a perfect match to your data. Also, from a anatomical point of view, most of the joint are constrained (e.g., knee translations are small).

[Mohammadreza Rezaie]

Thank you so much Dimitar,

Well, what about limiting the translations instead of locking them? e.g. 0.01 m RoM.

And how can we distinguish or get a better approximation of joint laxity from noise and skin artifacts? If this is not possible, the 6DOF model might not be a good idea.

Please let me know more about these.

Kind Regards,
Mohammadreza

[Dimitar Stanev]

Hi,

I think even if you limit the translations, the problem is not solved. Some degrees of freedom are very sensitive to the noise and it cannot be solved if you do not improve the measurement. Also, by unlocking all the degrees they can influence the results of the origin degrees of freedom.

[Mohammadreza Rezaie]

Thanks Dimitar,

I appreciate your help.