Power wheelchair driving analysis for people with motor disabilities using ANN classification

The present paper aims primarily to analyze the driving skills of patients with different pathologies using artificial neural networks. The evaluation of these patients' abilities to drive an electric wheelchair or power wheelchair shows that this battery-operated device can be dangerous for them when they present severe motor deficiencies. It is for this reason that it was deemed necessary to use a Power Wheelchair (PW) driving simulator in order to analyze, in an objective manner, their driving abilities. Consequently, an experimental study was carried out using the driving simulator at the Center for the Physical Medicine and Rehabilitation of Children (Centre de Médecine Physique et de Réadaptation pour Enfants) in Flavigny-sur-Moselle in France. The error between the reference trajectories and the patient's calculated trajectories which was used as input for classification allowed obtaining the model for analyzing the driver's skills. This model was then used for identifying the familiarized and novice users. The evolution of the above-mentioned error turned out to be an important indicator for improving the quality of the patient's driving skills during the learning phase.