Use this URL to cite or link to this record in EThOS:
Title: The production and control of functional electrical stimulation swing-through gait
Author: Heller, Benjamin Wolf
ISNI:       0000 0001 3552 3432
Awarding Body: University of Strathclyde
Current Institution: University of Strathclyde
Date of Award: 1992
Availability of Full Text:
Access from EThOS:
Access from Institution:
This thesis addresses some of the issues involved in the synthesis of swingthrough gait by functional electrical stimulation (FES). A general introduction is given to paraplegic gait, then the following areas are reviewed in detail: previous production of FES swing-through gait; biomechanical and energetics analyses of swing-through gait; general techniques for controlling FES gait; and the use of machine-learning techniques. Trained, non-impaired subjects wearing adjustable braces are used to model the movement patterns of FES swing-through gait. It is found that flexing the knees during the body-swing phase of swing-through gait reduces the energy cost of the gait. Hardware and software are developed to allow the production of FES swing-through gait in paraplegics with mid and low thoracic lesions of the spinal cord. The kinematic parameters of the resulting gait are assessed. It is found that the gait is faster than both knee-ankle-foot-orthosis (non FES) gait and reciprocal FES gait. This constitutes the first demonstration of FES free-knee swing-through gait in a spinal cord injured population. A symbolic inductive learning program, Empiric, is described. This program uses 'fuzzy' weighting to cope with uncertainty in the training data. This technique is found to offer improved classification performance (on artificially generated data) over both the orthodox (non-weighted) approach and an alternative weighting strategy. The fuzzy inductive learning technique is compared with a multi-layer perceptron type neural network for identifying the invariants (rules) that describe muscle activation during normal human gait. Both techniques are found to successfully model the muscular activation; the inductive learning technique has the advantage of producing explicit rules which are easily understood. The fuzzy inductive learning technique is applied to data obtained from the (previously mentioned) model of swing-through gait, in an attempt to mimic the control strategies used by the unimpaired subjects. It is found that the gait is best modelled with simple rule-sets, based on only one sensor. It is argued that this technique allows the automatic derivation of control strategies for FES gait: in particular, it allows the subjects' movement intentions to be determined. It is suggested that this 'intention detection' provides a more natural interface between a paraplegic subject and an FES control system than the techniques which are currently used.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
Keywords: Paraplegia; Biomechanics; Rehabilitation