Selective Assist Strategy by Using Lightweight Carbon Frame Exoskeleton Robot

Exoskeleton robots need to always actively assist the user’s movements otherwise robot just becomes a heavy load for user. However, estimating diversified movement intentions in daily life is not easy and no algorithm so far has achieved that level of estimation. In this study, we rather focus on assisting limited number selected by using an EMG-based classification newly developed lightweight exoskeleton robot. Our knee composed carbon fiber frame highly backdrivable joint driven pneumatic artificial muscle. Thus, our does interfere with motions even when actuator activated. As method, adopted positive-unlabeled (PU) classifier. Since precisely labeling all data from large-scale practical, assumed only part was labeled used PU classifier can handle unlabeled data. To validate approach, conducted experiments five healthy subjects selectively sit-to-stand four possible motions. We compared approach two methods assume fully The results showed subject’s were properly assisted.