My understanding about the kinematic input to CMC is that if the desired kinematics have already been filtered, such as being filtered in RRA step, it does not need to be filtered again in CMC step.
After I ran CMC on a walking trial with kinematics filtered (at 6Hz) in RRA, the residual forces are very jaggy. For example, the residual force applied at the pelvis in Y (green) and the pelvis acceleration in Y (blue) from a CMC simulation of walking are shown below. We see that both curves are very jaggy.
- The residual force applied at the pelvis in Y (green) and the pelvis acceleration in Y (blue) from a CMC simulation of walking.
- Residual and acceleration in Y.jpg (81.52 KiB) Viewed 255 times
The comparison of the acceleration of the pelvis in X (light blue), Y (purple) and Z (yellow) after RRA (dashed) and CMC (solid) from the same trial is shown below, in which the accelerations from RRA are smooth, whereas those from CMC are jaggy.
- The comparison of the accelerations of the pelvis after RRA and CMC.
- Pelvis Y acceleration CMC vs RRA.jpg (61.69 KiB) Viewed 255 times
The accelerations of pelvis in X, Y and Z directions look better when filtering (at 6Hz) is applied in CMC. The light blue, purple and yellow are translational acceleration in X, Y and Z directions, respectively.
- The accelerations of pelvis in X, Y and Z directions when filtering (at 6Hz) was applied in CMC.
- Pelvis acceleration after filtering in CMC.jpg (51.69 KiB) Viewed 255 times
Any thoughts on why this happened? Or, do we still need to filter the kinematics again before CMC? Thanks!