search

DanielFNG / motion-data-processing

Perform the necessary conversions and calculations from raw treadmill/Vicon data to allow for motion reconstruction in OpenSim
MIT License
0 stars 0 forks source link

Problems with GRF data (& potentially markers if from segmentation) #6

Closed DanielFNG closed 5 years ago

DanielFNG commented 5 years ago

Plot the x component of the CoP. There are spikes in two locations: just before it goes to 0, and after (at the end of the gait cycle). Think this is due to the filtering & segmentation process. Investigate potential fixes & changes to this.

DanielFNG commented 5 years ago

cop_fixes

Implemented CoP processing. Before & after shown above.

DanielFNG commented 5 years ago

Before in blue, after in red.

Fixed in cd526d46b34ae5614b8335bdd73db25715398752.

DanielFNG commented 5 years ago

Better not to extrapolate all the way down to 0. This is not natural when you look at the final data and if CoP's are inaccurate at low forces you can't assume this. Instead, we say that below the threshold at which CoP is accurate, we assume the Cop is equal to its value at the boundary during stance, and is undefined (or set to 0 for OpenSim processing purposes) outside of stance.

DanielFNG commented 5 years ago

Modifications to this fix implemented in commit cfa462a2f26b0b4360345e56d24a93f655443487