A wearable embedded platform for real-time hand neuroprostheses control

The control of upper limb neuroprostheses through the peripheral nervous system (PNS) can allow restoring motor functions in amputees. Beyond having a system, which could manage the prosthetic device with a satisfactory power efficiency, another crucial aspect is the real-time implementation of neural decoding algorithms, which necessarily need to run on a custom embedded system. This aspect is usually overlooked, notwithstanding the actual portability and the impact that limited hardware resources have on the efficiency/effectiveness of the decoding algorithms. This paper presents an embedded system that could allow controlling a neural hand prosthesis by decoding in real-time the movement intention extracted from the PNS activity. Interfacing this system with the robotic hand and an analogue front-end for signal acquisition and neural stimulation, it is possible to implement a wearable solution for the neuroprosthesis control. The final embodiment demonstrates the real-time performance onto a low-power off-the-shelf Digital Signal Processor (DSP), opening to experiments exploiting the efferent signals to control a hand neuroprosthesis