Model takes off during RRA

[Daniel Kohle]

Hello all,
I tried RRA with 200k and 500k integration steps. I also increased output precision and decreased the optimizer derivative and convergence criterion value. This didn't solve my problem.
Daniel

[Daniel Kohle]

The solution:
I finally managed to run RRA without the problems I described. The solution is to switch the kv and kp parameters higher. In my "standard" file it was 100 and 20. Now, with 1600 and 80, it works fine. I will see, whether I can complete CMC with these setting, too.

Daniel

[Giordano Valente]

Hi,
I had similar issues on RRA/CMC computing and I partially fixed the problem increasing the amount of residuals actuators in both RRA and CMC. The results show the RRA output kinematics not modified, but the CMC one doesn't track the recorded kinematics. My objective is to obtain muscles forces results without changing my original kinematics. Any suggestions?
Thanks.
Giordano.

[Joseph Alston]

Is it ever a problem for the optimal forces to be too large for RRA?

[daniel jaramillo]

Hi Dan,

Were you able to do the static optimization also?

Thanks

[Anders Sandholm]

Hi

Have you solved this?
I have found the exact program in some test crouch motion captures.

Does anyone know where this can come from? I also increased the kp and kv

cheers

[J.J. Mordang]

Hi,


Currently I'm busy with a model of the right leg. I've added 3 more joints to the foot to see what happens during normal gait. I wanted to do a RRA simulation and I read the rraCmcPage and the suggestions in an other thread but the simulation still won't work properly. When the RRA simulation is finished the result is that the model is transformed along the x,y and Z axis as one rigid body and no joint angles have been modified. Even after different kv, kp or force settings.
The Inverse Kinematics and the inverse dynamics look just fine.
Can anyone suggest what I'm doing wrong?

Thank you very much for your time and help!


Kind regards,

[Ayman Habib]

Hi J. J.,

Since it sounds you're modeling only one leg, can you point out what kind of experimental data you're using and what part of the gait cycle you're simulating? Stock file from the distribution (e.g. gait2354, 2392) have forces, residual-actuators for both legs. Also why run RRA, does InverseDynamics solution suggest some discrepancy between grf and the model-dynamics that you're trying to fix?

Maybe you can step back and reconsider the methodology to use to answer your research question.

Good Luck,
-Ayman

[J.J. Mordang]

Hi Ayman,


The main objective of my research project is to extract EMG data from the muscles in the lower limb and compare them with real data acquired in the lab. The main interests are in the foot bones and the tibialis posterior and anterior. However, there is only one force plate in our gait lab and all camera's are focused on that such that only the forces during normal walking can be measured in the right leg and foot (no treadmill option).
To make a model that satisfies the number of segments I wanted the foot to have I edited the Lower Limb Model 2009 provided by Arnold et al. I added 4 more joints to foot between the parts of interest of the footbones.
I read the document "generateing a Muscle-Actuated Simulation in OpenSim" by Anderson et al where they use the gait2354model to explain how it works. I also read the user guide very thoroughly especially the chapters about scaling, IK, RRA,CMC and FD. The .xml files needed for the RRA (setup/tasks etcetera) are based on the the gaitmodel by looking at the configuration (joints etcetara) of these xml files and modify them to my own model. The IK and ID look just fine, but I thought, because I use the document as a recipe, that RRA was needed. Because for the CMC the RRA_kinematics.sto is needed. Also when I skip the RRA and use the kinematics from the ID the CMC stops after 7%. I'm still looking if I can make the RRA better by playing with the setup but so far I have not succeeded.

To make a long story short. The only thing I wanted to do is to compare the EMG data calculated by OpenSim with real EMG data by using a segmented foot.

Thank you very much for your help Ayman, hopefully it's possible for me to fulfil my research.


Kind regards,

Jan-Jurre

[J.J. Mordang]

Hi Ayman,


I managed to do the RRA simulation by locking the pelvis and converting my lab data into treadmill data. However, when I want to compute muscle control I encounter a lot of problems.

1. When I check the "append to muscle's force data" instead of replace, the program just stops after printing : Setting cmc controller to not use verbose printing.

2. When I check the replace box, most times the CMC stops at a certain percentage (both slow and fast target). And if it finishes the model joints angles are all tangled up. I got one simulation nearly right (very high optimal force at the foot joints) but no muscle excitations could be seen after simulation.

3. I added 4 segments to the foot as that is what I'm interested in so I prefer not to lock them.

Are there any tips you can give me about configuring the .xml files needed for the CMC?

Thank you very much for your time and help.


Kind regards,

Jan-Jurre