Nonlinear Modeling and System Identification for Cortical Control of Arm Prosthetics

A nonlinear model was established based on ANN for the closed-loop system, which simulates the non-human primates' cortical control of a computer cursor's movement in a 3D virtual reality environment. Elaborately designed experiments were performed on monkeys who have chronically implanted microelectrode arrays in motor cortex areas corresponding to the arm. Monkeys were trained to control the movement of a cursor using cortical signals Using advanced system identification tools, the mappings between recorded monkeys’ arm movement parameters and neuronal activities in cortex areas were extracted. This approach improved substantially the accuracy in predicting arm trajectories from recorded cortical signals than reported in the literature using other methodologies. This new algorithm contributes to the promising application of assisting paralysed people to control neuroprosthetic devices.