Dear Opensim users and developers,
I would like to calculate the glenohumeral joint reaction forces during different type of tennis serves. To that aim I will use an opensim model (ground, thorax, clavicle, scapula, arm, forearm and hand) in which I have implemented a no-dislocation constraint of the glenohumeral joint. Thereafter I was thinking running static optimization and joint reaction algorithms to obtain the glenohumeral joint reactions forces.
My problem is that I did not measure the ground reaction forces nor the force between the racket and the hand. Thus I was wondering if I can still perform a static optimization without external forces. My hope is that the residuals (error) may be “gathered” at the dof between the ground and the thorax.
Yours sincerely
Yoann
Joint reaction forces without external force measurements
-
yoann blache
- Posts: 14
- Joined: Tue Oct 08, 2013 2:11 pm
Re: Joint reaction forces without external force measurement
A BallJoint, a GimbalJoint or any CustomJoint with rotational degrees of freedom will not dislocate. Can you explain what you mean by "implemented a no-dislocation constraint of the glenohumeral joint"? OpenSim can report Joint reaction forces for any joint type so it is generally unnecessary to employ constraints.
You can run static optimization and then the joint reaction analysis (applying the forces from SO) without external forces, but you will not get realistic loads at the shoulder without at least modeling the racket mass and inertia during the swing. Your model seems to be missing the tennis racket!
During the drive, the forces imparted on the ground to accelerate the body will be important to achieve the observed body and racket kinematics (how are those being measured/estimated?). Force-plate measured GRFs tend to be the most accurate measurement of center-of-mass acceleration so they would serve to improve the quality of the simulation. If you assume the Thorax is free floating (no trunk, pelvis and legs connecting to ground) you can use residuals to actuate its degrees-of-freedom but you will have no way to verify if those residuals forces are reasonable without measured GRFs.
During impact the glenohumeral joint reaction forces will be almost meaningless without at least estimating the load on the racket head or transmitted to the hand from the racket.
You can run static optimization and then the joint reaction analysis (applying the forces from SO) without external forces, but you will not get realistic loads at the shoulder without at least modeling the racket mass and inertia during the swing. Your model seems to be missing the tennis racket!
During the drive, the forces imparted on the ground to accelerate the body will be important to achieve the observed body and racket kinematics (how are those being measured/estimated?). Force-plate measured GRFs tend to be the most accurate measurement of center-of-mass acceleration so they would serve to improve the quality of the simulation. If you assume the Thorax is free floating (no trunk, pelvis and legs connecting to ground) you can use residuals to actuate its degrees-of-freedom but you will have no way to verify if those residuals forces are reasonable without measured GRFs.
During impact the glenohumeral joint reaction forces will be almost meaningless without at least estimating the load on the racket head or transmitted to the hand from the racket.
-
yoann blache
- Posts: 14
- Joined: Tue Oct 08, 2013 2:11 pm
Re: Joint reaction forces without external force measurement
Thank you very much for these remarks.
Concerning the model, I forgot to mention that we have implemented the racket in our model.
I agree that the estimated muscle forces are meaningless at the impact; that is why we often stop the analysis just before the impact.
Concerning the no-dislocation constraint at the glenohumeral joint (GH). Indeed if GH is modeled with three dof (3 rotations - ballJoint for instance), no dislocation will visually occur. Nevertheless if we look at the GH joint reaction forces, we observe that for some movements (usually fast or with high load) the ratio between shear and compression GH forces is superior to the ratio of GH dislocation calculated from cadaveric specimens (Lippit et al. 1993, Dickerson et al. 2007). Thus we implemented eight linear constraints (inequality) that enable to constrain GH reactions forces to be inside a friction polygon that respect the no-dislocation ratios. As mentioned by Bosterlee et al. (2013), we believe that implementing this no-dislocation constraint help to implement a more realistic simulation.
Best regards
Concerning the model, I forgot to mention that we have implemented the racket in our model.
I agree that the estimated muscle forces are meaningless at the impact; that is why we often stop the analysis just before the impact.
Concerning the no-dislocation constraint at the glenohumeral joint (GH). Indeed if GH is modeled with three dof (3 rotations - ballJoint for instance), no dislocation will visually occur. Nevertheless if we look at the GH joint reaction forces, we observe that for some movements (usually fast or with high load) the ratio between shear and compression GH forces is superior to the ratio of GH dislocation calculated from cadaveric specimens (Lippit et al. 1993, Dickerson et al. 2007). Thus we implemented eight linear constraints (inequality) that enable to constrain GH reactions forces to be inside a friction polygon that respect the no-dislocation ratios. As mentioned by Bosterlee et al. (2013), we believe that implementing this no-dislocation constraint help to implement a more realistic simulation.
Best regards