Pelvis Rotation

[Edilson Borba]

Hello

I have done some sessions where I calibrate the person facing the camera and perform the walk with my back to the cameras, this due to a space limitation in my laboratory. I realized that the rotation of the pelvis changes drastically and I understood that if I calibrate from the front and collect from the back the pelvis should rotate 180 degrees. For example, when collecting from the front the pelvis is rotated by +5 degrees and when collecting from the back it rotates between 175 and 185 degrees. Up to that point things seem normal, but what I can't understand is why in some cases the rotation value is positive and in others negative, for example, for the same session in different trials I have pelvic rotation around 180 degrees positive for one trial and around 180 degrees negative for another. I've made adjustments to the code to ignore this in the analysis, but I'd like to understand why this happens.



Thank you very much

[Matt Petrucci]

Hi Edilson,

It's simply the nature of the inverse kinematics being calculated.

The positive x-direction (or 0 degrees) would be in the direction of a vector pointing out of the calibration board. Since you are turned around in relation to that vector (walking in the negative x-direction), it's going to be values that are somewhere around +/- 180 degrees.

Inverse kinematics are solved going forward in time, so given you are either starting at 177 or -178 (which means the pelvis starts just 2 or 3 degrees off the x-axis), and the model is allowed to go past 180 or -180, each time step is going to follow a value closer to the previous time step. So for example, if your previous time step is -178, the next could be -179 or 181, but it's more likely going to use -179 because it is closer to -178.

Hope this helps