Abstract
A prototype pedaled wheelchair/ergometer system has been designed and developed for leg paralysis patients as part of the development of a rehabilitation system based on a brain-machine interface (BMI) for patient with motor paralysis. The wheelchair/ergometer system consists of a customized pedal driven wheelchair/ergometer and a BMI algorithm that detects the patient's intention. The patient's desire to move his (or her) legs is recognized from electroencephalography using current source density estimation and a support vector machine classifier. These recognition results are used for autonomous driving of the wheelchair/ergometer. The patient's feet are fixed to the pedals so that they move with the rotation of the wheels (pedals). Preliminary testing using this prototype system with an unimpaired subject showed that the discrimination accuracy between 'pedaling state' and 'resting state' was over 80% even in real-time operation.
