Asymmetrical muscle activation in exampleMocoInverse

[Karthick Ganesan]

Hi all,
When I ran the exampleMocoInverse code, I see some muscle activations are highly asymmetrical with full muscle activation (for example see the pic below). What is the reason for this and how to overcome this?

gaslat.png (13.27 KiB) Viewed 901 times

Thanks,
Karthick.

[Karthick Ganesan]

The following seems to be the reason, as I saw in another post by Nick.

"Without a periodicity goal or some sort of initial bound on the initial muscle activations, they are free to activate as much as they want at the initial time point. The activation state variable at the initial time is basically unconstrained because there is no previous time point for the activation dynamics model to limit it's value. With a periodicity constraint, you avoid this issue by using the end of the trajectory as the "previous" time point."

I tried different things like minimizing sum of squared activations, initial bounds and periodicity goal. None of them have worked.
Any suggestions will be helpful.
Thanks,
Karthick.

[Nicholas Bianco]

HI Karthick,

If minimizing activations squared doesn't work (try using a large weight), then it might just be that the joint moments at that time point are large, and the solver decided to use the glmin to meet the constraint for whatever reason. Also, the gait cycle should be roughly symmetrical, but they won't be perfectly symmetric.

It's possible that there's an artifact in the ground reaction forces near the initial time point, causing the spike in muscle activation.

[Karthick Ganesan]

Thanks Nick. I will try with larger weight.
Originally I was doing MocoInverse in Matlab with the model and data from the Moco Paper (Prescribed walking example), For which, I got similar kind of full activations for gaslat, bfsh,sart, glmin, psoas and tfl. I am wondering if you also got similar activations or different. In the paper among these muscles only bfsh appears. I am not sure the plot is for the left leg or average of right and left. Because in left leg, no such kind of increased activation is present.

reserve knee.png (16.4 KiB) Viewed 820 times

grf.png (20.31 KiB) Viewed 820 times

gaslat.png (20.42 KiB) Viewed 820 times

knee moment.png (18.66 KiB) Viewed 820 times

In the above plot for excitation, dashed line is left leg and solid line is right leg. The gait cycle is from 0.83 to 2 sec.
Thanks,
Karthick.