Motor Learning for Control of Prosthetic Limbs

In this paper we present results from five real-time myoelectric control experiments with able-bodied and amputee participants comprising three distinct paradigms, namely, two-dimensional cursor position control on a screen; classification-based grasp control of a prosthetic hand; and prosthetic control of individual fingers. We observe a consistent trend across this spectrum of myoelectric interfaces, that is, regardless of the level of intuitiveness of the interface, users can improve their performance in biofeedback myoelectric control tasks with training. We argue that progress in prosthetic limb control can be best achieved through a strong synergy of signal processing algorithms, motor learning, and feedback.