RRA Workflow

[Vishal R]

Hi,

I am trying to figure the "proper/recommended" workflow in using RRA to proceed with a model suitable for performing CMC, SO and forward dynamics. I am following the user guide https://simtk-confluence.stanford.edu/d ... d+with+RRA but a still confused with how to proceed with gait2392 model level ground walking analysis.

I did the following:
1. Scaled the model - checked with the checklist as mentioned in the user guide
2. Performed IK, ID - satisfactory results obtained
3. Before doing CMC, I ran RRA. As a result, it generated a new COM adjusted osim file with mass recommendations.
The result looks as follows:

Code: Select all

************************************************************
*      Summary of Mass Adjustments to Reduce Residuals     *
************************************************************
* Body adjusted: torso
* Mass Center (COM) adjustment: dx =0.0268045, dz =-0.00730106
* New COM location: ~[-0.0616796,0.372002,0.00730106]
************************************************************
* Recommended mass adjustments:                             
*  Total mass change: -7.29711
*  pelvis: orig mass = 14.1641, new mass = 13.0208
*  femur_r: orig mass = 11.1867, new mass = 10.2837
*  tibia_r: orig mass = 4.45899, new mass = 4.09906
*  talus_r: orig mass = 0.120269, new mass = 0.110561
*  calcn_r: orig mass = 1.50337, new mass = 1.38202
*  toes_r: orig mass = 0.260503, new mass = 0.239476
*  femur_l: orig mass = 11.1867, new mass = 10.2837
*  tibia_l: orig mass = 4.45899, new mass = 4.09906
*  talus_l: orig mass = 0.120269, new mass = 0.110561
*  calcn_l: orig mass = 1.50337, new mass = 1.38202
*  toes_l: orig mass = 0.260503, new mass = 0.239476
*  torso: orig mass = 41.1761, new mass = 37.8524
************************************************************
* Note: Edit the model to make recommended adjustments to  *
*       mass properties.                                   *
************************************************************
* Average residuals before adjusting torso COM:
*  FX=-16.8099 FY=71.5602 FZ=-8.06263
*  MX=2.94817 MY=-1.979 MZ=10.8236
************************************************************
* Average residuals after adjusting torso COM:
*  FX=-16.9376 FY=71.6655 FZ=-8.54885
*  MX=-1.15003 MY=-2.5351 MZ=0.18652
************************************************************

************************************************************
*                   Final Average Residuals                *
************************************************************
* After torso COM and Kinematics adjustments:
*  FX=-1.20729 FY=1.65838 FZ=-1.6221
*  MX=-2.97286 MY=-2.53186 MZ=-0.736567
************************************************************

4. Should I edit the new osim file with mass recommendations? And further, do I have to run RRA again on the adjusted model? What values do I have to look for, Average residuals before/after adjusting torso or Final Average Residuals?

How would I find the end to this 3-4-3 iteration loop? Only after this, I could do CMC and other analysis on the new adjusted model. If someone could share their experience it would be great!
Thanks a lot.

Kind regards,
Vishal

[Thomas Uchida]

You may wish to refer to the "The Strength of Simulation: Estimating Leg Muscle Forces in Stance and Swing" example in the Confluence documentation (https://simtk-confluence.stanford.edu/d ... +and+Swing). See, in particular, "Part II: Leg Muscle Force Estimation in Stance" (https://simtk-confluence.stanford.edu/d ... +in+Stance), Sections B and C, which walk through a typical RRA workflow.

[Vishal R]

You may wish to refer to the "The Strength of Simulation: Estimating Leg Muscle Forces in Stance and Swing" example in the Confluence documentation (https://simtk-confluence.stanford.edu/d ... +and+Swing). See, in particular, "Part II: Leg Muscle Force Estimation in Stance" (https://simtk-confluence.stanford.edu/d ... +in+Stance), Sections B and C, which walk through a typical RRA workflow.

Thank you for pointing to the tutorial. I followed them, but I couldn't check my answers to the questions asked in the tutorials.

I have been trying to study the stance phase(normal level ground walking), and after few iterations of RRA, I got the following results:

Code: Select all

************************************************************
* Average residuals before adjusting torso COM:
*  FX=-17.3697 FY=-0.713774 FZ=-5.80722
*  MX=0.0242314 MY=-2.65218 MZ=0.0099284
************************************************************
* Average residuals after adjusting torso COM:
*  FX=-17.3706 FY=-0.712469 FZ=-5.80762
*  MX=-0.00160677 MY=-2.65279 MZ=0.00181453
************************************************************

************************************************************
*                   Final Average Residuals                *
************************************************************
* After torso COM and Kinematics adjustments:
*  FX=-3.45099 FY=-22.4073 FZ=-5.02137
*  MX=-2.05297 MY=-2.76582 MZ=-2.81319
************************************************************

The mass recommendations shown now is ~0.07 Kg. And I have adjusted the tracking task weights for pelvis_tx,pelvis_ty and pelvis_tz from 5 to 25,100 and 25 respectively. Now, In the GUI I can see the simulation without any drifting and flying. The thing that I dont understand is how to get the results with respect to the table RRA Guide

I tried plotting the RRA_actuation_force.sto as shown below:


RRA_controls.sto


I have set the optimal force for Fx, Fy and Fz as 6,20 and 4N respectively in the RRA_Actuators.

if Max residual force is considered from the Actuation_force.sto, then the results are very bad. and if I consider the average residuals, they are in negative. Could you please explain how to bring the residual forces in the good or OK range.

Thank you.

Kind regards,
Vishal

[Shivangi Giri]

Hello Vishal,
I have been trying to understand working of RRA and how to optimize the weights to reduce residuals.
So far I have not had much luck. I went through your post and thought you might have gotten a better understanding of RRA now.
After running RRA for several iterations the pelvic moment residuals do not seem to reduce but increase at every iteration and the values are large. Also there is drift in pelvic residuals.
Kindly suggest what I can do.

Thanks and Regards,
Shivangi Giri

[Vishal R]

After running RRA for several iterations the pelvic moment residuals do not seem to reduce but increase at every iteration and the values are large. Also there is drift in pelvic residuals.

Hello Shivangi,
I am not aware of what activity are you trying to analyze in your study. So assuming it as a level ground walking as discussed in my original question, I could mention a few pointers while doing RRA:

• After each iteration of RRA, you may have to edit the mass of the subject manually and re-run using the new model file with any changes mentioned in the COM from the output.

• Have a look at the ground reaction forces(especially the time where reaction forces are 0). Your RRA is only valid if you have proper GRF for both the limbs(for the time window you are studying). If not you may have to trim the simulation time(both the limbs cannot be in the air/no GRF forces while performing RRA, this will trigger reserve forces to compensate experimental data). Most of the problem is solved by focusing on this

• Do not add huge residual forces for reserve actuators Fx, Fy, and Fz. They would result in an un-realistic simulation output. If you are just doing simulations for walking, the table mentioned under RRA in the documentation stays valid. These values would differ if you are working on activities other than LGW.

Hope this helps! Good luck,
Vishal

[Shivangi Giri]

Thank you for your reply.
I am analyzing ground level running for a healthy adult. Yes, I change mass of each segment after each RRA iteration.
Also, I have put weights on tasks I want to track closely and have gone through the RRA troubleshooting and best practices, but all in vain. The optimal force value for all pelvic residuals is 1 N and 1 Nm for forces and moments respectively.
GRF force data for both legs is available, but the alignment is somewhat not accurate. I have explored all the extraction tools available on OpenSIM but the forces I get are either not aligned or do not have COP information.
Could you please tell me the best tool to extract OpenSIM compatible GRF data from c3d file?
Thanks in advance.

Regards,
Shivangi

[wish lock]

HOw can we get any study visa for KSA. I want to go Saudi and want help regard Iqama fee and their fine. Can anyone help me.