simtk.org

Hello,

I am currently investigating a running pattern, where the subject is running over a single force plate.
I performed IK, ID of the stance phase (where I am interested in) without any major problems and got similar results as in literature. I am using the Lai_Arnold_2017 model (refined from Rajagopal, all models have RMS <0.01)).

I perform RRA in MATLAB, where I made a custom made autmatic process of RRA iterations. When I am performing RRA to reduce my residuals, I noticed that after a few iterations the recommended mass adjustment was converging to zero (although I did not change any masses of segments) and I noticed that the change of the CoM did not much in reducing residuals. What was left is the changed Kinematics_q; which was considerably reducing my residuals. When I compared Kinematics_q with the IK result, it was actually the same however the Kinematics_q had a lag of about 0.01 sec.

The result of RRA (from _Actuation_force.sto) was that my Fx, Fy and Fz residuals where very close to zero (<4 N), however my Mx,My,Mz were considerably larger (see attachment), but just inside the boundaries from the wiki. The next step for me was to perform SO with the adjusted Kinematics_q file. Here, I got Fx, Fy, Fz residuals very close to zero again, but my Mx, My, Mz residuals again higher (even higher as the result showed in RRA).

In addition, the reserve actuators of the hip_rotation, ankle and subtalar joint were really high compared to the other joints (which were close to zero) (see attachment SO_result). If seems that after sample 100 something strange is happening, but I can't see some weird things in my model. My muscle forces from the SO results almost all go directly to an activation of 1.0 during the movement. Does somebody have a tip what to do to get a proper result from my SO?
Is there something wrong with my IK? Is my model not correct?

Hi Robbert,

How fast were the participants running across the force plate? Do all muscles go to an activation of 1.0 or is it, for example, the gastroc. M. and L or soleus (assuming those are in the model)? It could be that the muscles in your model might be "weak" for your task. Have you tried increasing the maximum force generating capacity of these muscles and still get this activation of 1.0?

All the best,

Mike

Hi Michael,

Thanks for your reply!

The participants were running at approximately 12 km/h over the force plate. Not all muscles are going to a maximal activation, but these do at least:
Addbrev_r, addlong_r, Illiacus_r, Piri_r ,Psoas_r, Recfem_r, Sart_r,Tfl_r
Ehl_l, Fdl_l, Fhl_l, Gas_lat_l, Gasmed_l, Glmed_l, Glmin_l, Illiacus_l
Piri_l, Recfem_l, Soleus_l, Tfl_l, Tibiant_l, Tibpost_l, Vasint_l, Vaslat_140, Vasmed_l

I think that is has something to do with 'weird' kinematics around sample 100 (that is why the actuators become really active of hip adduction, ankle and subtalar for example). Maybe the 'weird' kinematics assure that muscles are not on their optimal length during that part of running and therefore reach an activation of 1. When I check the maximal force of each muscle divided by their maximal isometric force, a lot of the muscles do not reach an activation of 1 here, but in the SO_activation.sto file they do.

But I don't see very weird kinematics around this time stamp; at most some hip sagging but that is also visible in the VICON Nexus data.

Increasing the maximal isometric forces of all muscles (times 2), does help to get the activations below 1.0 for almost all (except fhl_r, gaslat_r, glmax3_r, tibant_r) but I don't know if this is justified.

Do you think it has something to do with the kinematics? Or something else?

Hope to hear from you,

Robbert