I have a question regarding the sensitivity of a FDK-controlled measurement in respect to the initial definition of the measurement constrain.
The problem is the following:
I am currently working on the lumbar spine model of the AAUHuman model with the aim to control the intervertebral joints by FDK. To do so, I customized the lumber Rhythm (Driver SpineRhythmDrv) in JointsLumbar.any by defining CType as forcedep while driving a rotational measure between Pelvis and T12 to make the model fully constrained.
This worked without problem and the InverseDynamics results were also reasonable, thus there seem to be no error in the implementation of the FDK control.
As FDK solves the kinematics stepwise by finding the model configuration which minimizes the supplementary forces that balance the system, the result should be independent from how the joint configuration was defined initially.
However, when I subsequently edited the entries of SRMatrixes.any in the last 3 columns on the right hand side the result of the Study changed remarkably. If I for example all non-zero entries on the the right-hand sight in the first three rows are multiplied by 10 (i.e. if the rotation of the L5-Sacrum joint is assumed to be 10 times the rotation in the T12-L1 joint) the result of the inverse dynamics changes by up to 10 % compared to default. Thus, the inital definition of the configuration of the FDK-controlled measure seems to notably influence the outcome of the FDK derived force equilibrium.
Does anybody had the same observation for FDK controlled joints?
What could be an explanation for this behavior?
Is there a best practice for the selection of initial values (i.e. measurment drivers) for FDK controlled joints?
Thanks for any answers!