Whole-body model for shoulder kinetics/kinematics


Dear AnyBody,

Is it possible to make minimal changes to the existing MoCap_FullBody model, to simulate simple upper-body tasks such as grasping/moving small objects? I don’t foresee any problems in calculating kinematics, but not sure what to do with the external forces, which is from GRFs measurements if this is a walking/running study. Is it necessary to replace forceplates with another form of force-sensing device?

Thank you,


Hi Phobe,

In terms of GRF predictions the same code should be possible to reuse, but in one or the other way the forces in the hands needs to be applied. If it’s a small object being moved it would be simplest to measure also the position of this object, drive, assign mass properties, and connect the hand to it kinetically.

If the subject is exerting forces in other ways you can create contact forces like on the feet’s, or alternatively measure the forces using transducers and apply them in the hands.

Hope it helps

Best regards


Dear Søren,

Thank you for your reply!

The object is small and light-weighted, so I think in terms of experiments, we should put markers on the object and the upper extremity (shoulder/elbow/upper and lower arms, etc.), define/measure the trajectories of all markers, and then use the object’s position to drive the upper-body model. Is that correct?

As the full-body MoCap model also predicts JRFs in the upper body, I am wondering if these two approaches provide similar results: 1. an upper-body model driven by the positions of an object connected to the hand (as you suggested); and 2. a full-body MoCap model with GRFs measured (GRFs may vary little with time, as the subject sits still to perform the task).

Thank you!


Hi Phobe,

I would make model having both arms and object driven by mocap. Then add reactions between object and hand and apply correct mass and inertia properties of the object.

If the object is touching anything except the hand it is a different story?

Concerning the two different approaches i would use the first one assuming the object has no contact with anything except hands. The models would in the ideal world give same results, but experimental errors on makers and difference in subject mass distribution and model mass distribution will make results different.

Best regards