A shared control policy for center-out movement decoding in motor Brain-machine Interface

Brain-Machine Interface provides a new way to control the peripheral devices directly using signals from brain. However, because of the uncertainty and instability of brain signals, the decoding method cannot fulfill the demand of accurate control of the intended movement. We proposed a shared control policy to involve environmental information into the decoding process of brain signals. While the monkey manipulated the joystick in a center-out task, the trajectory was updated with a control signal that derived from current decoded kinematic information considering the potential targets. Our results showed that using the proposed method combined with the decoding process, the correlation coefficient between the predicted trajectory and the true signals increased by 17.4% in average. It indicated the control policy involved the environmental information could greatly improve the performance of motor brain machine interfaces in practice.