Humans use multi-objective control to regulate lateral foot placement when walking.

Area of Science:

• Human motor neuroscience
• Computational modeling
• Biomechanics

Background:

• Human walking involves complex motor control to manage physiological noise and redundancy.
• Existing bipedal walking models struggle to replicate human equifinality and dynamic balance.
• Understanding how the nervous system regulates stepping movements is crucial for developing accurate computational models.

Purpose of the Study:

• To extend the Goal Equivalent Manifold (GEM) framework for predictive modeling of human walking.
• To test hypotheses regarding how humans regulate stepping movements to achieve specific goals (lateral position, speed, step width).
• To compare model predictions with human experimental data to identify key control strategies.

Main Methods:

• Developed computational models based on the Goal Equivalent Manifold (GEM) framework.
• Modeled stepping movement regulation for single and combined goals: maintaining lateral position, lateral speed, and step width.
• Compared model predictions against experimental data from human walking.

Main Results:

• Uni-objective models showed redundancy but failed to replicate human stepping dynamics.
• Most multi-objective models balancing two goals also failed to match human data.
• Multi-objective models prioritizing step width regulation over lateral position successfully replicated human step-to-step dynamics.

Conclusions:

• Human lateral stepping movement regulation is inherently multi-objective, involving trade-offs between competing goals.
• Prioritizing step width regulation is key to replicating observed human walking dynamics.
• GEM models serve as 'control templates' that complement mechanical templates for understanding human locomotion.