Optimal time-varying postural control in a single-link neuromechanical model with feedback latencies

Area of Science:

• Biomedical Engineering
• Neuroscience
• Robotics

Background:

• Maintaining balance during quiet standing is complex for neural control due to inherent task instabilities.
• Sensory feedback delays and muscle low-pass properties challenge real-time postural regulation.
• Inverted-pendulum (IP) robotic models are common for studying postural balance control.

Purpose of the Study:

• To develop an in-depth neuromechanical postural control model based on physiological principles.
• To propose an optimal proportional-integral-derivative (PID) controller for effective postural control despite sensory feedback latencies.
• To assess the efficacy of a time-varying PID controller tuned with linear quadratic regulator (LQR) principles.

Main Methods:

• Developed a single-segment IP robotic model incorporating a Hill-type muscle model.
• Included proprioceptive feedback from muscle spindle (MS) and Golgi tendon organ (GTO).
• Employed computer simulations and sensitivity analysis to evaluate the PID-LQR controller's performance.

Main Results:

• The tuned PID-LQR controller demonstrated effective postural stabilization.
• Sensitivity analysis revealed a delay tolerance of up to 300ms for the controlled system.
• Model predictions of center of mass (COM) excursion correlated highly with empirical data from perturbation experiments.

Conclusions:

• The developed neuromechanical model provides a robust framework for understanding postural control.
• The proposed PID-LQR controller effectively manages sensory feedback latencies, crucial for dynamic balance.
• The model's high correlation with empirical data validates its physiological relevance and predictive capability.