Research On Brain Machine Interface Control System Based On Motor Imagery

Brain Computer Interface(BCI)is a technology that establishes direct communication and control channels between a human or animal's brain and a computer or other electronic devices instead of using the normal output channels of the brain-peripheral nerves and muscles.By collecting and recording the brain's EEG signals,specific neuronal activity in the brain can be identified by some technical means such as signal processing,pattern recognition,and machine learning.Then it can output some simple instructions for communicating with the outside world or controlling external objects.In the early stage of BCI study,BCI technology was used primarily in the medical field,helping the motor function and sensory deficits to restore basic motor and sensory function which improve their quality of lives greatly.Nowadays,BCI is used as a research tool and widely used in education,military,games and entertainment.Typically,the principle of a BCI application is firstly to collect brain activity of the brain electrical signals,then through a series of algorithm processing,and finally output control signals for external object control.The BCI system can be divided into three categories according to whether it invades the brain: invasive BCI,partially invasive BCI,and non-invasive BCI.The motor imagery based BCI in this paper is non-invasive,a classifier is trained by the subject's EEG of motor imagery and later used for online brain activity recognition and classifying.In a BCI system,the processing of EEG signals is the key factor,including the removal of noise signals,feature extraction and classification of unsteady EEG signals.The accuracy of the algorithm recognition determines the control effect and the user experience directly.In this paper,the key technologies of motor imagery based brain-computer interface is studied,including: 1)experiment design and acquisition method of motor imagery EEG signal;2)signal preprocessing techniques,multivariate empirical mode decomposition algorithm used for artifact removal and result comparison withindependent component analysis(ICA);3)feature extraction and classification of EEG signals are studied.The common spatial patterns(CSP)and linear discriminant analysis(LDA)is applied to BCI Competition dataset and dataset collected by our equipment;4)two online brain machine interface wireless control system is built based on the above studies,realization of motor imagery based BCI with cursor control and wireless car control.The framework of the two systems,hardware and software implementation and working principle are described in detail.Based on the above research,I will conduct further research to improve and optimize the algorithm and control,preparing for the subsequent research of controlling the exoskeleton manipulator based on brain-computer interface.