Question about computing wrist joint moment

[Yihui Zhao]

Hi,
I want to get the wrist joint moment. flexion/extension and ulnar/deviation. And I have stuck for months. I

Briefly, I used VICON to capture the wrist motion and convert to trc files. I followed the tutorial in OpenSim and use the Upper limb dynamic model (MoBl-ARMS Dynamic Upper Limb) for Inverse kinematic and inverse dynamic. The Inverse Kinematics results showed around 0.005 total squared errors. I locked some coordinated, which matched the experiment posture and free the wrist flexion/Extension and Ulnar/Radial deviation.
The problem is the inverse dynamics results changes sharply and didn't show the expected results.
In my opinion, The curve of wrist flexion moment show decrease when from neutral position to flexion and increase from flexion to neutral position.
Please correct me if I am wrong.
And Any ideas to solve this problem?
Many Thanks