Power of Assist Control based on Wearer's Forearm using Motion Estimation Motion Sensor

Powered exoskeletons are becoming popular especially in fields of nursing care and agriculture. We have been developing a powered exoskeleton that is supposed to be used in case of a nuclear hazard. Conventional powered exoskeletons typically use electromyographs (EMG), force switches, or force sensors on the wear's feet. Unfortunately, the former suffers from the decline in measuring accuracy or slips of EMG sensor by sweating. The force-switch-based assist control method has a difficulty in speedy assist control of walk because it needs a few steps of a walk for the recognition. The force-sensorbased assist control method realizes a rapid assist control but the assistive motion is defined beforehand in general. We propose a new approach for power assist controller based on user motion prediction using 9 axis motion sensor. The motion sensor detects the angle and angular velocity of the wear's limb and estimates the appropriate angle of the powered attachment according to the estimated motion. This report conducts experiments with one degree of freedom powered arm to evaluate the proposed method by EMG sensors and a force sensor