Assessment of residual reduction procedures for high-speed tasks

Area of Science:

• Biomechanics
• Computational modeling
• Human movement analysis

Background:

• Experimental and modeling errors often cause dynamic inconsistencies in inverse dynamics analyses of human movement.
• Residual pelvis actuators can mitigate these errors, but may remain large in demanding tasks, necessitating further strategies.

Purpose of the Study:

• To investigate if an optimal control algorithm can adjust measured kinematics during the preparatory phase of a high-speed, torque-demanding task (single leg triple hop) to maintain low residual forces and moments.

Main Methods:

• An optimal control algorithm was formulated as a tracking problem using implicit dynamics.
• Equations of motion and residual pelvis forces/moments were incorporated as path constraints.
• GPOPS-II and IPOPT were used for optimization, with OpenSim API for inverse dynamics calculations.

Main Results:

• A significant reduction in all six residual actuator components was observed throughout the analyzed task.
• Optimized motion closely matched reference motion before the ascending phase; some discrepancies emerged in pelvis tilt and lumbar extension during this phase.

Conclusions:

• The proposed algorithm effectively addresses dynamic inconsistency in high-speed human movement analysis, achieving low residuals while preserving kinematic similarity.
• This approach can serve as a valuable complement to existing optimal control methods for generating dynamically consistent simulated motions.