Difference between model based on GRF Prediction and real force plates data

Hi, I use mocap example model for my simulation of some lifting tasks. I use both GRF prediction model and Full body model which is based on my measured force plate data. The problem I am having is that in some of the asymmetric postures with high rotations, the result I get for L5-S1 forces is very different between two models while my setup and marker protocol is the same. In addition the L5-S1 forces that Full body model (which is based on measured GRF data) shows is much higher than other model while the predicted GRFs and COPS are close to measured data from lab. I want to know why the L5-S1 forces' results for some of these asymmetric postures with high rotations between two models is so different while I do not have any issue with symmetric postures or even asymmetry postures with small rotations.

Hi @Sajjad - and welcome to the forum!

  • Do you have the same trials with GRF prediction and force plate data ? or is it different trials doing the same motion?

    • how much does the kinematics differ between the trials?
  • When you say "high rotations" what does that mean?

Best regards,

Dear Bjorn, thanks for your response,
I have a problem with lifting tasks where the subject lifts the load from his frontal plane. For example, the load is on the right side of his body , therefore there is a high rotation and lateral bending in the subject. I believe my problem comes from muscle overloading because I get an overloading warning. However I do not know why despite the overloading the model based on GRF prediction shows lower forces for spine loads. I use the same trials and everything is the same. In other tasks where overloading does not occur, the amount for spine loads is the same for both models but when muscle overloading happens the model based on force plates show higher values. The Kinematics for both models is the same. Does the GRF prediction model have a mechanism which controls the effect of muscle overloading?

Hi @Sajjad

The GRF prediction models use a quadratic recruitment criteria whereas the force plate models use the standard cubic criteria - this is to add numerical stability since the artificial muscles in the GRF prediction have a large strength difference compared with the residual forces on the model.

If your movement has a large degree of trunk rotation that is probably why you see overloaded muscles. The thorax is rigid so large degrees of rotation might resolve in some muscles getting very small moment arms.

When the muscles are overloaded and the warning is issued the simulation will continue but the solver will switch to the min/max criteria in order to switch off the upper bound on the muscle activity.

Best regards,

This topic was automatically closed 125 days after the last reply. New replies are no longer allowed.