Control at stability's edge minimizes energetic costs: expert stick balancing

Area of Science:

• Motor control
• Human sensorimotor systems
• Robotics

Background:

• Stick balancing is a complex voluntary motor task requiring stabilization of an unstable system.
• Expert human performance involves specific parameters: 0.23 s time delay, 0.32 m minimum stick length for 240 s balance, and a defined dead zone for angle estimation.

Purpose of the Study:

• To develop a computational model explaining expert stick balancing.
• To investigate the role of internal models in compensating for time delays in motor control.

Main Methods:

• A switching-type, pendulum-cart model was developed to simulate balance control.
• Numerical simulations were performed using the semi-discretization method.
• Analysis focused on feedback gains tuned near the edge of stability.

Main Results:

• The model incorporates an internal model to predict sensory consequences and compensate for time delays.
• Simulations indicate feedback gains are optimized near the stability limit.
• A cost function considering fingertip position and corrective forces is minimized.

Conclusions:

• Expert stick balancing relies on a predictive internal model to manage time delays.
• Optimal control strategies balance maneuverability with energy efficiency.
• The findings provide insights into human sensorimotor control and robotic applications.