question regarding SO

[Michał Wołek]

Hello

I am doing static optimization and the results have sudden jumps for all muscles at once. During computing there weren't any errors and constraint violations were very low all the time. Here are the results for glut_med1,2,3 muscles: http://prntscr.com/pxtuz5
This is for sum of muscle activations =2 and using muscle force-length-velocity relation

What can I do to improve results?

[jimmy d]

If you haven't yet, perhaps have a look at that Static Optimization tutorial , it discusses jagged activations. There also could be discontinuities in either your kinematics or, if the jumps are around heel strike or toe-off, due to the forces (thus accelerations) changing suddenly. There have been previous forum posts that showed that filtering. Perhaps filtering the kinematics or running the data through RRA will help.

[Michał Wołek]

I have both filtered kinematics and I am using RRA states for kinematics. I tried increasing optimal forces on pelvis point and torque actuators and the jumps are smaller, but still occur. Also more residual forces equals less natural outputs. Or should I not care and just increase residuals even more?

Also what "performance" value is considered as being acceptable? if I get on average 1-1.5 is is rather good or too much?

[jimmy d]

I would suggest posting a zip of all the files needed to recreate the issue. It could be poor kinematics, forces, model scaling, muscle scaling.

[Michał Wołek]

Yeah, I considered bad scaling/too little foot markers. I have problems with putting markers on feet in a way they behave more naturally during inverse kinematics. But I have no movie of subject during recording so I cant compare. Forces are measured on forceplates, they should be quite good.

https://drive.google.com/open?id=1XbRxU ... VKollAgGq7