Muscle-actuated simulation error

[Jon Akins]

I am having a similar problem detailed in the 'RRA and CMC' post below. I successfully followed the 'Generating a Muscle-Actuated Simulation in OpenSim' and wanted to increase the duration of the simulation. XML Notepad was used to edit all .xml files.

1) Changed the time range from '0.4 1.6' to '0.4 4.6' in the subject01_setup_ik.xml and successfully ran the ik module.

2) Changed the Final Time from '1.25' to '4.25' in the subject01_setup_RRA.xml and successfully ran the CMC module using the edited file.

3) Changed the Final Time from '1.2' to '4.2' in the subject01_setup_CMC.xml and ran the CMC module. The program consistently crashes without warning at approximately 11%.

4) I was able to successfully increase the duration of the simulation from a final time of '1.2' to '1.7'.

5) Changed the Final Time from '1.18' to '1.68' in the subject01_setup_Forward.xml and successfully ran the forward module with the edited file. However, the ankle and knee distort rapidly at the end of the forward motion file (~1.5 s).

I plan to adjust muscle activation magnitudes/durations and compare the resulting knee angle at heel strike with the original model.

Your help is much appreciated,

Jon Akins

[Daniel Leib]

Hi Jon,

Have you tried altering the integrator settings to "loosen" it up a bit? There are a number of kinematic simulations I have tried that simply won't run for long due to disagreement (I'm pretty sure anyway) between experimental and model data, despite the fact that they tend to line up pretty well visually. I imagine it may get even tougher when you add forces to the simulation. Different trials of the same person collected the same way will run fine, however. Go figure! If any of the OpenSim team have a solution or ideas as to what the problem might be for this, I'd be all ears.

I've tried to use more of the demo files as you're trying to do as well but had the same problem. It can go longer if you allow for a noisy output, though.

Have a good one,

Dan

[Jon Akins]

I am able to complete CMC and Forward modules when using 2.0 to 3.0s with some model distortion (left ankle and knee) near the end of the simulation.

Altering the muscle excitations is my next challenge. I adjusted excitations of a few muscles in the subject01_walk1_control.xml file located in the ResultsCMC folder. After running the Forward module the simulation completes, but the model moves erratically.

Do I also need to adjust the corresponding activations in subject01_walk1_states.sto?

[Ayman Habib]

Hi Jon,

It's great that you were able to run CMC and Forward to get muscle excitations.

Forward runs in Open-loop (there's no feedback to tell the model to adjust if it's off its originally tracked motion). Arbitrary muscle excitations will definitely make for erratic model movements. It's actually quite hard to come up with an excitation pattern that causes coordinated movement.

The file _states.sto is an output file so you shouldn't need to change it.

Good Luck,
-Ayman

[Jon Akins]

Just to clarify...I was referring to the _state.sto output file from the CMC results which is used as an input to the Forward module.