kinematics for external loads

[Ryan Bakker]

Hello,
Could somebody please explain what “kinematics for external loads” in the GRF file does?

The documentation (for ID) states “Next, you can specify the Kinematics for external loads file. In general, you should leave this blank. OpenSim will then use the specified kinematics that are being tracked by ID, RRA, and CMC (e.g. for ID for RRA it would be IK kinematics, and for CMC it would be RRA kinematics).

When would you want to use a specified kinematics file that is different from the tracked kinematics?

Also, when I run RRA or CMC with and without the specified kinematics file I get slightly different results. If I use the same specified kinematics that I am tracking, should they be the same?

-Ryan

[Ryan Bakker]

To give an example of the differences I am seeing, I have included two sets of figures.

The first two figures are RRA pErr results. One with the "kinematics for external loads" and one without. The rest of the parameter files are exactly the same.

The second set of figures are CMC reserve actuator results. One with the "kinematics for external loads" and one without. The rest of the parameter files are exactly the same.

As we can see there is a large difference between the two results. Should they be the same?

Kinematics for external loads.pdf

(256.81 KiB) Downloaded 121 times

[Ajay Seth]

Kinematics for external loads enables external forces to be registered and expressed in the body the force is being applied to. Why do we do this? In Inverse Dynamics this has absolutely no effect since the kinematics are set by the input kinematics (say from IK) so that when the equivalent point of force application on the foot (e.g. the center of pressure on the ground) is determined it is based on the current configuration of the model which is set by the input motion. In RRA, it is solving a tracking problem via forward simulation and it permits deviation from the target kinematics (this is the adjust kinematics option for reducing residuals) so that the location of the foot (w.r.t. ground) can vary dramatically. There is one camp that says the forces should be described (registered) with respect to the foot based on the input (IK) kinematics to tolerate more drift in the forward simulation. The other camp says no, the kinematic adjustments should keep the foot in more or less the correct global position and that the external forces should remain in the lab frame and applied to the equivalent point on the foot based on the actual position of the model. I tend to be in the second camp, because it preserves your measurement of the location of the center of pressure and requires adjustments to the model kinematics to still satisfy that measurement. It is also far less confusing to me. In some cases, (like yours) registering the external force in the foot frame according to IK kinematics does tolerate more drift in the kinematics, but that might not be a good thing. I hope that explains the differences you are seeing. To improve the quality of the results you may have to track the kinematics more closely so that it does not cause the foot to deviate too much (without kinematics for external loads).
It is confusing.

[Ryan Bakker]

Ajay,

Thank you for your detailed post. It was very informative and helpful. I think I have a good grasp on it now.

Does anybody know which method Sam used in his most recent paper "Muscle contributions to fore-aft and vertical body mass center accelerations over a range of running speeds"? I have downloaded some of his old files and it seems that Kinematics for External loads were used. Can anybody confirm that this method was used in the most recent paper?

Thank you
-Ryan