simtk.org

Dear OpenSim community!

I'm using OpenSim now for some month already but I'm still struggeling with the RRA pipeline to put the result into the CMC and forward dynamics tool. IK, ID and static optimisation are working. I'm using the gait2392 musculoskeletal model.

What I want to do is a muscle force prediction during normal, healthy walking. I'm using VICON to measure kinematic and GRF data during one gait cycle (two steps), and take the raw data as imput into OpenSim. I've already captured data for three subejcts, for the first one RRA is working fine, for the second and the third one I've got problems while using RRA.

For the second subject I get following error after a while:

"OPTIMIZATION FAILED...

CMC.computeControls: WARN- The optimizer could not find a solution at time = 1.481000.
If using the fast target, try using the slow target.
Starting at a slightly different initial time may also help."

I tried using the slow target and at different initial time steps but it did not work. I know that I could change the muscle bounds to make them stronger but I want to change as less as possible and because I have a lot of trials it would be quite time consuming to do that. Does someone has another solution for this problem?

For the third subject, I can run the RRA tool properly but the walking model is sinking into the ground after the first walking step, so it fullfills another step but sinks as there would be no floor anymore. (I tried to upload a video but I got following message "Sorry, the board attachement quota has been reached".)

Why does this happen? In gait 2392 there is a "ground" inserted which you can see in the GUI. The sinking of the model happens right after the model passes this ground so I wonder if this has something to do with it? But then the RRA tool worked fine with my first subject and this did not happen. Did someone experienced something like this as well?

I found different muscle models (Thelen2003, Millard2012 acceleration/equilibrium) and I'm thinking about trying the Arnold2010 model as well. Could that solve these problems?

I would be glad for any help!

Nice greetings from the UK
Ursula

What I forgot to explain, when the RRA stops and gives me the Error message output, it is often that before it stops the model does a weard ankle turn in the transversal plane. It starts to walk normaly then after a while the only movement can be seen in the ankle.

Hm...I have the feeling that these errors result of the GRFs I have. First, I use 4 force plates and in the _grf.xml file there were only two force plates defined, so I defined my other two froce plates and the ID tool worked. But now RRA crashed again after some time frames. When I use "preview experimental data" the result seem OK.

Can someone help me out here? Can I only fix this error through making the muscles stronger?

Thank you!!!

OK problem solved.

First, we figured out that there were some errors with our force plates (ant-post and med-lat directions were not right). Then the ...grf.xml file which is used for the subject_ID.xml file was not right set up and we changed that through the GUI so everything works fine now! Still some small sinking into the ground (around the hight of the feet) but we suspect that this is because of other errors of the force plates.

Hi Ursula
I have the same problem with RRA and CMC. I'm thinking of the file format so what program you used to convert your vicon files to opensim trc and mot file (I'm using lee-son toolbox). also how you create your grf.xml file.
thanks
abeer

Abeer,

I use the Lee-Son toolbox as well and it seems to work fine. The _grf.xml file you can create when you browse the subject_ID.xml file (open the Inverse Dynamics Tool) and then go on "external loads". There on the very right click on the button next to the search folder button. Then browse your trial_grf.mot file and define the forces. Then safe and you have your _grf.xml file. Here is the description I followed:

http://simtk-confluence.stanford.edu:80 ... Tutorial+3

Hope that helps!

Hi Ursula
Thanks a lot for your replay. I just want to double check with you, when you view your static and dynamic trc and mot files in GUI, does your model stand on the top of the plate (if you apply gait2392 model) because when I view for example my static trc in GUI all the markers seem fine but my trc model stand just outside the plate and the same with my mot file, the model walks also outside the plate. regarding the grf.xml file, I'm using the same method to create the file, we are using 6 force plates so the feet strike 3 force pltes per trial which means that I have 3 external forces. first, I figure from the mot file which force plate has been striked by which foot (right or left foot) and based on this info I start create my EXT forces, for example if the right foot strike plate 1, left foot strike plate 3, and right foot strike plate 5, then for my first external force (Right_GRF) I check "ground_force_vx1" , "ground_force_px1", and "ground_torque_x1". for second external force (Left_GRF) I check ground_force_vx3" , "ground_force_px3", and "ground_torque_x3". for third external force (Right_GRF) I check ground_force_vx5" , "ground_force_px5", and "ground_torque_x5". however I still have a sinking model when I apply RRA and CMC.
thanks
abeer

To be honest, sometimes my model stands slightly next to the force plate as well when applying the scaling and IK tool, but I don't know why this occurs and if this has an impact on the results. I hoped it is only an error in the visualisation.

Second, after applying the RRA tool, my model still sinks slightly into the ground (around the height of the feet of the model) but when finishing the full gait cycle it is on the right level again. As the first force plate lets the model sink and the second one brings it up again. We were discussing if it has something to do with the wrong body weight of the model but this should be right. We had some issues with the force plates so I assumed that this error occurred because of that. I have not done any new measurements till now so I cannot confirm this theory.

Could someone else help us out here? Does it matter when the model in the GUI stands slightly next to the force plates? Or is that only a visualisation problem? Or has it something to do with the Lee-Son toolbox? And why is my model still slightly sinking into the ground? It happens to nearly all my trials and both subjects I already measured.

Thank you a lot!!!

Hi Ursula
I just post my issues in the public forums, hopflly some one will respond. however you did not tell me if the way I created my grf.xml file is correct or not.
thanks
abeer

Hi,

All body geometry in OpenSim (except for contact and wrapping) is used exclusively for visualization. I'd worry if the ground reaction force vectors look off or not lined up with the model but not about where things are relative to the plane that we display as ground.

Hope this helps,
-Ayman