Encouraging stroke survivors to actively participate in robot aided gait training is critical for optimizing the outcome of this intervention. In this respect, it is of crucial importance that the timing of the provided assistance and the amount of assistance is in accordance with the subjects needs. We tested the feasibility of a control algorithm for a powered exoskeleton that selectively supports foot clearance and adapts its support to the performance of the subject. This was done in five chronic stroke survivors with stiff knee gait. Foot clearance was selectively supported through a virtual spring between the desired and actual ankle height of the paretic leg. The virtual spring stiffness was automatically adapted based on the experienced movement error in the previous step and a forgetting factor. The results showed that the virtual spring stiffness profile converged to a steady state pattern in about 20 steps. The pattern was subject specific and was roughly shaped to the deviation of the actual ankle trajectory from the reference trajectory before the assistance was turned on. The assistance resulted in an increased foot clearance through increased knee flexion, whereas it left other aspects of gait unaffected. The presented algorithm turned out to be effective in providing appropriately timed assistance according to the subjects needs.
|Conference||World Congress on Medical Physics and Biomedical Engineering 2009|
|Period||7/09/09 → 12/09/09|