simtk.org

Hi there,

I'm simulating squatting using my own motion capture data, but when I'm running inverse kinematics I found the foot is not flat(the toes is downward). I think the foot should be flat and the preview marker data also shows the foot is flat when squatting. How should I reduce this error? In the scalling step, the error has been reduced to under 2cm. How should I reduce the error in IK and keep the foot is flat?

I have tried to change the coordinate value(ankle_angle_r/l) and fix it to ensure the foot is flat, but it will cause another problem.
Should I change the parent_frame of the markers on foot(eg. FOOT_R_LAT) to ground and fix it?

Any help will be appreciated.

Thanks,
Yanran

Hi, yanran

I meet the same problem with you. I find that this phenomenon comes from the errror of hip and knee joint angles. I have tried to adjust the position of the virtual marker but it has no effective changes. Waiting for professional reply.

Simon

Hi Yanran & Simon.

One thing you could try is create virtual markers at the 'floor' of your participant based on the static trial. I've done this more-so for scaling purposes by projecting markers that are on the feet down to the floor (i.e. y=0). You could use a similar process and also keep these virtual markers in your dynamic trials - and subsequently they would sort of represent the bottom of the foot. If you heavily weighted these virtual markers for IK I think it would do a reasonable job of keeping the feet positioned appropriately on the floor relative to your participants movement.

Aaron

Hi Aaron,

Thanks for your reply. I partially understand your meaning. Could you provide a model that contains your setting?

Best wishes,
Simon

Hi Simon,

I can't really provide a model - I typically add these markers as a part of the processing steps in my code. The general gist is (for the static trial at least) that you add a new marker to the .trc file below any markers you have on the foot, where you take the X and Z data from the original marker but just make the Y constantly zero. This needs to correspond to a marker in your .osim file that sits on the foot at the ground level.

In thinking about this further, a more simple solution if you want the foot to stay in contact with the ground is to have a weld joint between the foot and ground bodies. That may not represent experimental data appropriately given the foot likely lifts off the ground a little - but if you are determined to have it completely flat on the ground a weld would do it.

Aaron

Hi Aaron,

Thanks a lot. Now the feet are flat.

Best wishes,
Simon