Modeling of control and learning in a stepping motion

Area of Science:

• Biomechanics
• Robotics
• Neuroscience

Background:

• Previous work modeled stepping with a simplified three-body linkage and open-loop control.
• Hierarchical control structures were suggested for normal stepping operations.

Purpose of the Study:

• To refine a dynamic model of human stepping by incorporating post-landing leg and foot dynamics.
• To improve simulation techniques using cycloidal preprogramming for complete step motion.
• To propose a hierarchical control model with learning for efficient movement sequencing.

Main Methods:

• Refined dynamic model including leg and foot segments.
• Lagrangian method to derive ground forces and torques during landing.
• Simulation compared against experimental data from obstacle stepping using Selspot motion analysis.
• Novel curve-fitting procedure to test cycloidal velocity profile hypothesis.

Main Results:

• Simulation results align with experimental data for obstacle stepping.
• The proposed hierarchical control model efficiently combines open and closed-loop strategies.
• Evidence suggests multiobjective optimization in joint motion control.
• Cycloidal velocity profiles were supported by experimental data.

Conclusions:

• The refined model and simulation techniques enhance understanding of stepping dynamics.
• The proposed control and learning model offers insights into assembling movement segments for goal-directed actions.
• This work contributes to understanding human motor control and learning mechanisms.