Fig. 6

Panel a shows a schematic of the proposed control architecture. Myoelectric signals collected from 8 sEMG sensors were processed using linear discriminant analysis (LDA) classifiers and grouped in 5 classes: no movement, hand open, power grasp, fine pinch and index point. These classes were used to control the define grips of the SoftHand 2 Pro through the simultaneous activation of the 2 DoA. Panel b-d show the experimental setup used by amputee subjects: b shows hardware components including 8 sEMG sensors (1), the customized liner (2) and socket (3), a support interface compatible with the Ottobock quick disconnect wrist (4) and the SoftHand 2 Pro prototype (5); c presents an example of surface electrodes distribution on an amputee subject; d shows the system worn by a user during the experimental session