Experimental performance evaluation of human balance control models

Thomas P. Huryn, Jean-Sébastien Blouin, Elizabeth A. Croft, Michael S. Koehle, H. F. Machiel Van Der Loos

Research output: Contribution to journalArticleResearchpeer-review

8 Citations (Scopus)


Two factors commonly differentiate proposed balance control models for quiet human standing: 1) intermittent muscle activation and 2) prediction that overcomes sensorimotor time delays. In this experiment we assessed the viability and performance of intermittent activation and prediction in a balance control loop that included the neuromuscular dynamics of human calf muscles. Muscles were driven by functional electrical stimulation (FES). The performance of the different controllers was compared based on sway patterns and mechanical effort required to balance a human body load on a robotic balance simulator. All evaluated controllers balanced subjects with and without a neural block applied to their common peroneal and tibial nerves, showing that the models can produce stable balance in the absence of natural activation. Intermittent activation required less stimulation energy than continuous control but predisposed the system to increased sway. Relative to intermittent control, continuous control reproduced the sway size of natural standing better. Prediction was not necessary for stable balance control but did improve stability when control was intermittent, suggesting a possible benefit of a predictor for intermittent activation. Further application of intermittent activation and predictive control models may drive prolonged, stable FES-controlled standing that improves quality of life for people with balance impairments.

Original languageEnglish
Article number6804652
Pages (from-to)1115-1127
Number of pages13
JournalIEEE Transactions on Neural Systems and Rehabilitation Engineering
Issue number6
Publication statusPublished - 1 Nov 2014
Externally publishedYes


  • Balance control
  • control models
  • functional electrical stimulation
  • standing

Cite this