A human interface predicting intended strides for a paraplegic locomotor device

Gait reconstruction systems for paraplegics have been attempted to improve their activities of daily lives. The stride control based on the user's intension is important to coordinate the voluntary movements of the user and the assisted movements of paralyzed legs. Assuming that the intended stride is equal to the distance of the preceding an arm movement, we developed a human interface estimating the distance of the arm movement using an accelerometer. The locomotor device is controlled so that the stride length is equal to the arm movement distance. In the current work, we examine a human interface and a control system of a locomotor device. A hand distance is estimated through polynomial fitting of the arm acceleration based on the criterion of smoothness. Here, we report two experimental results; (1) evaluation of the estimation accuracy, (2) validation of the control system of the locomotor device. Comparing the estimation accuracy of polynomial fitting with that of a twice integral method, the estimation of the polynomial fitting method shows good consistency with the measured distance. In addition, we confirmed that the stirde length can be controlled through the voluntary arm acceleration. These results suggest that the proposed human interface is effective for a paraplegic locomotor device.