Keynote Lecturer: Dario Farina
Active upper limb prostheses are controlled via a man-machine interface that establishes a link between the user’s nervous system and the robotic limb. This interfacing is commonly done with the remnant muscles above the amputation. In commercial systems, the intensity of muscle activity is extracted from the EMG and used for the direct control of single degrees of freedom. This type of control is limited to 1-2 degrees of freedom. More advanced control systems have been proposed in academia, either based on signal classification into a finite set of classes, or on continuous mapping (regression) of the EMG signal into the multiple degrees of freedom space(simultaneous and proportional control of multiple degrees of freedom). We review these advances, especially those based on regression. Moreover, the exclusive use of EMG as a source for feed-forward control of prostheses may not be sufficient and methods that integrate the EMG information with that from other sensors, within semiautonomous systems, may be preferable in the future. The talk will cover these topics with a discussion on the major challenges in filling the gap between commercial/clinical and academic methods for myocontrol.