A universal closed-loop brain-machine interface framework design and its application to a joint prosthesis

Brain-machine interface (BMI) system offers the possibility for the brain communicating with external devices (such as prostheses) according to the electroencephalograms, but there are few BMI frameworks available for flexible design systems. In this paper, first of all, inspired by the single-joint information transmission (SJIT) model, a Wiener-filter-based decoder and an auxiliary controller based on model predictive control strategy are designed to rebuild the information pathways between the brain and prosthesis. Specifically, the decoder is used to decode the neuron activities from cerebral cortex, and the auxiliary controller is used to calculate control inputs, which injected to the SJIT model as feedback information. Then, a universal closed-loop BMI framework available for designing flexible systems is proposed and formulated on the basis of the brain model, decoder, auxiliary controller and prosthesis, and it can well recovery the motor function of prosthesis. Finally, a simulation and another experiment are designed to show that the presented closed-loop BMI framework is feasible and can track the target trajectory accurately, and the presented framework with actual prosthesis can successfully achieve the target position along the target trajectory.