Behavior Identification for Wheelchair Driver Using the Fuzzy C-Means Classification Method

This study deals with behavior identification of Electric-Powered Wheelchair (EPW) drivers. User’s actions on the joystick determine this behavior. We assume that a driver evaluation can be performed using reference behaviors which are determined a priori. This study was carried out with two valid populations. The first one includes eight experimented subjects. The second includes six novice subjects. We propose to implement the Fuzzy C-Means (FCM) classification to separate different behaviors. The FCM was applied on the experimented population in order to determine reference’s behaviors. This shows that there were basically two different behaviors. An evaluation is performed on novice users by comparing their behaviors with respect to the reference ones. The results show that this population has an erratic behavior during the learning phase. Subsequently, users converge to one of the identified reference behaviors. We also evaluated a subject who suffers from muscular dystrophy and uses an EPW in everyday life. The results show a steady driving behavior.