CONICET | Buscador de Institutos y Recursos Humanos

Active lower limb assistivedevices are intended to contribute to human movement activities such aslocomotion. While some devices are capable to function among a variety ofterrains (i.e. level surfaces, inclines, or stairs), wide instrumentation sets(combining mechanical and bio-signal data)have been usually applied to minimize classification errors for locomotion moderecognition, even if this potentially bias user comfort. Hence, a selection ofparameters for classification with limited sensor data could represent amilestone on the development of comfortable lower-limb assistive devices. Theobjective of this study was to find the best combination of parameters for theoffline classification of five locomotion modes based on data from discreteareas on a pressure insole. Three virtual force-sensingresistor sensors were located under heel and forefoot regions of six subjects withoutdiscernible gait abnormalities to study the potential of kinetic signals forlocomotion-mode recognition. Parametric tests were used to compare the effectof two types of data, four time-windows and the combinations within five datafeatures on the classification error of linear discriminant analysisclassifiers for steady-state gait. Statistical comparisons were made using classificationerrors measured through five-fold cross-validation. Results showed that the best overall classifier had anaverage classification error of 7.85 4.76%.Ourresults were in close resemblance with related studies which relied on larger sensor sets as their instrumentation strategies. This methodpresents an insight about the use of a reduced number of sensors for accuratelocomotion mode classification on the development of comfortable lower limbassistive devices.

enviar mensaje