jump simulation with opensim 2.2.0

[Bruno Zaffaina]

For my thesis I have to do a simulation of a drop-landing (from a height=30cm).
I'm a beginner in this program and also in this analisys so I ask if anyone (experienced in this argument) could help me.
Now I use the model gait2392 and I'm able to do Scaling, InverseKinematics and InverseDynamics.
I can also do StaticOpt but the activations obtained aren't good.My goal is to get the muscle activations for comparison with EMG experimental signals.
Is there anyone who can help me?
thanks in advance,
bruno

[Ayman Habib]

Hi Bruno,

StaticOptimization does not utilize activation dynamics (no integration is done) and as such the results are not expected to be smooth. I'd switch to CMC which is a more elaborate and complicated process but would observe this better. Also in CMC you can define constraints to help guide the agorithm. If you go this route, however, you'll need to run RRA first.

Hope this helps and good luck,
-Ayman

[Bruno Zaffaina]

For me I have a problem but I don't understand which it is,the results of my ID on max pelvis residuals (absolute value) are the following:
pelvis_tilt_moment = 2800 Nm
pelvis_list_moment = 3100 Nm
pelvis_rotation_moment = 160 Nm
pelvis_tx_force = 5200 N
pelvis_ty_force = 9000 N
pelvis_tz_force = 4700 N
For my opinion these are too big.
But I don't understand where is the problem; any ideas?

[Bruno Zaffaina]

All of these max values are in the frame where there is the impact of the first foot on plate of force...
any ideas?

[Eric Greska]

Bruno,

I have successfully modeled this and had quite a few errors initially. The first item to consider is the fact that the models within OpenSim are set for gait analysis. With this thought in mind realize that the force generating capability of the model may not be ample to account for the ground rection forces developed at initial contact. My first successful import of the data had my subject contacting the ground and then getting blasted off as though they stepped onto a landline. This was due to the fact that the original model could not absorb the impact force, therefore the ground threw her up and off. By increasing the force capabilities of the model, you will also decrease the residuals that you were seeing. It took me quite a few trials of dialing in an ample amount of force for the actuators, but I can now pull in data from any subject, scale it, and put it through the fast cmc process completing in about 5 minutes.

Another thing to consider is the initial frame you are choosing. You state that it is at initial contact, but if the GRF vector is bouncing around, try to choose the initial point that the vector remains in a somewhat consistent pattern. I had a few trials that went from a vertical vector at IC, to a posterior, and then back to vertical, within each subsequent frame, and this would always cause a crash.

Let me know if you have other questions modeling the drop-jump.

Eric

[Bruno Zaffaina]

thanks for the answer!
I have some questions for you:
1) how much is your frequency cut-off for filtering data of markers and of GRF?
2) how do you choose muscles in the model to be increased the optimal force?
3) how much do you increase the optimal forces? x2?
I'm sorry for my english, but i hope you understand me!

[Ayman Habib]

Hi Bruno,

I expect to see some residuals on the first frame where ground reaction forces are applied but I don't expect them to be this big. This may suggest that ground reaction forces are not applied or not applied correctly (incorrect units or frames).

We usually filter at 6 hz. and it's your decision what muscles to increase and by how much depending on activity.

Hope this helps,
-Ayman

[Bruno Zaffaina]

Hi Ayman, the problem is the choice of freq. cutoff. If I set 7Hz, the new residuals of pelvis are:

pelvis_tilt_moment= 311.7392675 Nm
pelvis_list_moment= 118.6056695 Nm
pelvis_rotation_moment= 43.40200833 Nm
pelvis_tx_force= 329.4430604 N
pelvis_ty_force= 359.4548521 N
pelvis_tz_force= 39.33464002 N

Now I try to run the RRA analysis.
These residuals are ok or not?

[Bruno Zaffaina]

During RRA analysis the program runs until the frame of impact (t=2.58 s) and then it stops and crushes down with java error: http://imageshack.us/photo/my-images/83 ... emrra.gif/.
These frames are the same where the StaticOptimization doesn't run well.
I also try to increase <optimizer_max_iterations> from 2000 to 20000 but I have not solved anything.
I use the same file of gait2392 for:
RRA_Actuators
RRA_ControlConstraints
RRA_Tasks
Setup_RRA

But if I chage time range after the impact the RRA runs without errors.
Why?

Any ideas?

[Bruno Zaffaina]

Sorry but the RRA runs after the impact but the kinematics isn't well because the model collapses into the ground!
Why?