Related Experiment Video
Updated: Jun 16, 2025
07:19
A Modified Lean and Release Technique to Emphasize Response Inhibition and Action Selection in Reactive Balance
Published on: March 19, 2020
5.9K
Delayed reinforcement learning converges to intermittent control for human quiet stance.
Yongkun Zhao1, Balint K Hodossy1, Shibo Jing2
1Department of Bioengineering, Faculty of Engineering, Imperial College London, London, SW7 2AZ, United Kingdom.
Medical Engineering & Physics
|August 19, 2024
Summary
Human balance control may not be continuous. This study used a learning algorithm to model the human body, finding that intermittent neural signals, not constant ones, are optimal for maintaining upright posture.
Area of Science:
- Neuroscience
- Biophysics
- Control Theory
Background:
- The neural control of human quiet stance is debated, with differing views on the brain's role in maintaining upright posture.
- Classic models like continuous impedance control face instability issues when neural time delays are considered.
Purpose of the Study:
- To investigate optimal neural control strategies for human quiet stance using a computational model.
- To determine if continuous or intermittent neural activation is more effective for maintaining balance.
Main Methods:
- A one-segment inverted pendulum model representing the human body was used.
- A delayed reinforcement learning algorithm was developed to discover an optimal control policy without pre-imposed strategies.
Main Results:
- The developed algorithm identified an optimal neural controller exhibiting intermittent, rather than continuous, activation patterns.
- Simulation results suggest that intermittent control is a viable strategy for maintaining balance.
Conclusions:
- The findings support the intermittent control model for human quiet stance.
- The central nervous system may intermittently provide neural feedback torque to maintain upright posture, challenging continuous control theories.
