Back to search

University of British Columbia

Behavior shift to altered physics law of standing : a prediction from the reinforcement learning controller of postural control

Abstract

dc:description

A central question to our understanding of postural control is the overall goal of standing balance. Current opinions of the topic diverge: researchers have argued that minimization of movement variability or overall exerted torque could be potential goals of balance control. The purposes of the thesis were to (1) model standing balance control using the Markov Decision Process framework and identify best parameter combinations that represent the physiological characteristics of standing and (2) probe the goal of standing using computational simulations and a custom-designed robotic balancing platform with altered standing balance dynamics. Human standing balance in the anterior-posterior direction was modeled using the Markov Decision Process framework, and the Q-learning algorithm was applied to solve the control problem. Performance of the model was evaluated by comparing the range, root mean square, mean power frequency and 99% power bandwidth of the simulated center of mass data with empirical evidence. In the experimental study, participants (n = 3) were asked to balance on the robotic balancing platform during perturbations in which torque bias terms were added to the load-stiffness relationship of standing. The exerted torque and body angle were recorded and analyzed. The simulated quiet standing behavior from the Markov Decision Process model resembled the frequency characteristics of standing with larger variability in the time series analysis. In the experimental study, two participants balanced at a more backward (forward) angle when positive (negative) torque bias terms were added, which matched the predictions from my hypothesis. However, the size of the angle shifts differed from the hypothesis and they did not maintain the same torque level as my hypothesis which predicts participants would maintain their torque. In conclusion, the Markov Decision Process model generated behavior close to human balance control given specific parameters. While the direction of body angle shifts observed in the human data and Markov Decision Process model simulated data matched the prediction from my hypothesis of torque minimization, the experimental results did not fully support the statement that people always seek to maintain their torque levels during standing.

Degree

thesis:*
Name thesis:degree_name
Master of Science - MSc
Level thesis:degree_level
master's
Discipline thesis:degree_discipline
Kinesiology
Grantor dc:publisher
University of British Columbia
Year dc:date
2021

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Wang, Jiyu
Contributors dc:contributor
  • Blouin, Jean-Sébastien

Rights

dc:rights
Statement dc:rights
  • Attribution-NonCommercial-NoDerivatives 4.0 International
Language dc:language
eng

Identifiers

dc:identifier.*
Handle dc:identifier
http://hdl.handle.net/2429/80593

Chain of custody

source
Harvested from
University of British Columbia
Base URL
circle.library.ubc.ca/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Wang, Jiyu. Behavior shift to altered physics law of standing : a prediction from the reinforcement learning controller of postural control. master's thesis, University of British Columbia, 2021. http://hdl.handle.net/2429/80593