Exporting CoM trajectories

[HK Kim]

Hi

Is there anyway we can export whole-body CoM trajectories data (e.g CoM velocity and acceleration at every time frame) during dynamic movement from OpenSim?
For instance, after simulating our model, we want to see if there is any change of CoM afterward.
Any advice would be greatly appreciated.

Thanks
H

[Carmichael Ong]

The BodyKinematics analysis tool could be helpful for you. You can check out some documentation on analyses here: https://simtk-confluence.stanford.edu/d ... h+Analyses.

[HK Kim]

Thanks, one more question.
Are we also able to export 3d markers trajectories from the opensim?

[Carmichael Ong]

The PointKinematics analysis could be helpful here. You can look up the marker locations on the bodies, and then use those as inputs in the PointKinematics analysis tool.

[HK Kim]

Thanks for your response Carmichael.
We have modified the muscle property of the hip abductors in opensim and to see if the whole-body CoM is changed. However, we found that it's not changed at all. It looks like the OpenSim model locates a optimal position of the CoM even one of the muscle is significantly modified in the generic model. By any chance, can you explain me why this modification does not affect the CoM?
Thanks,

[Thomas Uchida]

What type of simulation are you running? If you are generating your results by tracking experimental data, then changing muscle properties will not change the trajectory of the model. You might want to look at the "Overview of OpenSim Workflows" page in the documentation (https://simtk-confluence.stanford.edu/d ... +Workflows) for descriptions of inverse vs. forward problems.

[HK Kim]

Thanks for your response.
We have used forward dynamics using the CMC results as input data (cmc_states.sto and cmc_controls.xml).
We have modified muscle property (e.g. reduced optimal force) and run the forward dynamics to see if the center of mass is changed.
However, the model looks very unstable (goes crazy) and also the center of mass almost does not change.
Are we using the right method?
Could you kindly give us any advice on this?

Thanks

[Thomas Uchida]

You started with kinematics and ran an inverse analysis to compute muscle forces. If you apply these muscle forces to the model in a forward dynamic simulation, you should not expect to see the same kinematics as what you started with. This is generally true for systems with inherent instability. Even very small differences (round-off error, integration time stepping, etc.) will lead to large deviations. We need to sense the world around us to maintain balance when we walk; your model needs a controller.