Kinematics different between RRA and CMC

[Madeline Grosklos]

Hi,

Does anyone know what might cause kinematics to change significantly (5 degrees RMS; 8 degrees max difference) between RRA and CMC? I am using the fast optimization target during CMC.

I am finding that ankle kinematics are consistently shifted 5-8 degrees in my CMC results even though kinematic results from RRA match the inverse kinematics solution well. I do not have this issue at any other joint, only the ankle. All other joints are within 2 degrees for IK solution, RRA kinematics, and CMC kinematics.

I am also finding that changing the strength (max isometric force) of the ankle plantar flexors in the model changes the CMC kinematics soultion for the ankle joint, but changing the strength of the hip flexors (iliopsoas) in the model does not change the CMC kinematics solution at the hip at all. I would not expect muscle maximum isometric force to change the kinematics solution from CMC.

I am using OpenSim 4.4. Thanks in advance for any thoughts or insights!

[Jon Stingel]

Hi Madeline,

Below is a guide on how CMC works and why you might be getting the results that you are seeing. Based on your description it sounds like the plantarflexor strength is affecting the model's ability to complete the motion that you are simulating. A few questions that might help:

• What motion are you simulating?

• Do you have reserves at the ankle, and are they highly active?

• If you continue to increase the strength of the plantarflexors does the angle discrepancy change?

If the muscles in the model are not strong enough to complete the motion, you would expect the reserves to activate, and perhaps the kinematics to deviate in the way that you are seeing.

Hope this helps!
Jon

https://opensimconfluence.atlassian.net ... +CMC+Works

[Madeline Grosklos]

Hi Jon,

I am only simulating walking gait. The reserves at the ankle are highly active, but I feel like this is a result of the kinematics showing more plantarflexion at push off than what is observed in motion capture.

What I think is most interesting is that, when the muscles at other joints are weakened significantly, it does not change the kinematics. For the knee and hip the reseves simply kick in to satisfy the torque that the muscles are not providing to keep the kinematics the same.

The only variable I have modified that helps keep the kinematics closer to IK and RRA and decrease the reserves is inceasing kp and kv terms in the CMC _Tasks file. Are you familiar with these terms and/or any potential downstream consequences of modifying them? They are not mentioned in the guide on how CMC works.

Thanks!
Maddie

[Madeline Grosklos]

I wanted to follow up on this in case anyone else comes across similar challenges and finds my original post.

After taking hints from these two previous posts:
viewtopicPhpbb.php?f=91&t=4037&p=9228&start=0&view=
viewtopicPhpbb.php?f=91&t=4126&p=0&start=0&view=

I experimented with changing the kp and kv gains, as well as decreasing the CMC look ahead window.

I found that increasing kp and kv for the ankle did help minimize the pErr and reserve torque at the ankle, however the improvement (when modeling with a lambda of -15; kp=225 kv=30) was still not enough to meet quality standards for all participants. I am also not sure how changing kp and kv for one joint only would impact the entire system.

Decreasing the CMC look ahead window from the 10ms default to 2ms significantly improved pErr at the ankle, and consequently decreased ankle reserves for all participants. I have not yet figured out what it is about my gait data that requires this decreased CMC look ahead window, but it seems as though (based on my experience and the posts linked above) if you are having trouble tracking ankle kinematics during CMC, decreasing the CMC look ahead window is a great place to start when troubleshooting.

Cheers,
Maddie