Chulalongkorn University researchers developed a Lagrangian inverse-dynamics model integrating thigh, shank and foot. Against OpenSim 4.0, walking estimates showed about 10% normalised RMSE across major joints, while squatting retained a notable magnitude offset.
Key findings
- Walking results agreed well with OpenSim at roughly 10% normalised RMSE. Squatting preserved temporal phase but showed a sizeable magnitude offset, and estimates were sensitive to COP trajectories, foot geometry and force orientation.
Why this matters globally
A simpler torque-estimation pipeline could support rehabilitation robotics and movement analysis where complex 3D modelling is impractical, provided task-specific accuracy boundaries are respected.
Thai researcher contribution
Chulalongkorn mechanical engineers developed the model and clarified interactions among foot geometry, COP and ground-reaction force that matter for exoskeleton control.
Limitations to consider
The model is two-dimensional and evaluated on benchmark datasets rather than clinical users. Squat offsets show task-dependent accuracy, and no closed-loop exoskeleton validation is reported.
Verify the original sources
MathematicsRead the original article↗DOI: 10.3390/math14132400