This paper proposes the cybernetic rehabilitation aid (CRA) based on the concept of direct teaching using tactile feedback with electromyography (EMG)-based motor skill evaluation. Evaluation and teaching of motor skills are two important aspects of rehabilitation training, and the CRA provides novel and effective solutions to potentially solve the difficulties inherent in these two processes within a single system. In order to evaluate motor skills, EMG signals measured from a patient are analyzed using a log-linearized Gaussian mixture network that can classify motion patterns and compute the degree of similarity between the patient's measured EMG patterns and the desired pattern provided by the therapist. Tactile stimulators are used to convey motion instructions from the therapist or the system to the patient, and a rehabilitation robot can also be integrated into the developed prototype to increase its rehabilitation capacity. A series of experiments performed using the developed prototype demonstrated that the CRA can work as a human-human, human-computer and human-machine system. The experimental results indicated that the healthy (able-bodied) subjects were able to follow the desired muscular contraction levels instructed by the therapist or the system and perform proper joint motion without relying on visual feedback.

-The evaluation and teaching of motor skills in relation to patients are two important aspects of motor skill training. These two points or problems must be resolved in order to make such training effective. To address the issues simultaneously within a single system, this study proposes a Cybernetic Rehabilitation Aid (CRA) under the concept of direct teaching using tactile feedback with an EMG-based motor skill evaluation function. The CRA involves a human-machinehuman (physiotherapist-rehabilitation robot-patient) interface known as a Cybernetic Interface Platform using biological signals not only to monitor patients' motor skills but also to directly teach such skills to them. The CIP can also be used as a human-human (physiotherapist-patient) system as well as a human-machine (physiotherapist-rehabilitation robot) system. In order to evaluate motor skills, the motions of the physiotherapist (T) and the patient (P) were analyzed using a loglinearized Gaussian mixture model that can classify motion patterns via electromyography (EMG) signals. Tactile stimulators were used to convey the instructions of the therapist or the system to the patients. A rehabilitation robot known as the Biodex System was integrated into the developed setup for a number of rehabilitation tasks.