A Study On Hybrid Brain-Computer Interface And Its Application

In this paper, we apply multiple degrees of freedom and asynchronous control-key technology in wheelchair control base on motor imagery brain-computer interface. The characteristics of wheelchair controlling are multi-degree-of-freedom and fast decisionmaking, which make brain-controlled wheelchair as a great challenging issue. To address this issue, firstly, we propose a sustained and brief motor imagery brain-computer interfaces controlled by two levels control mode, resulting in obtaining more controlling commands and greater information transfer rate. Secondly, in order to realize asynchronous control, a threshold mechanism has been applied for combination control to overcome error triggered defects. Thirdly, deep convolution neural network is applied in feature extraction of motor imagery EEG segments based on offline analysis to dig out more hidden features. Next, these features are combined with the artificial selected features to construct high quantity features for classification. So a novel and effective feature extraction and classification algorithm under a unified framework is proposed for EEG preprocessing.Firstly, we propose an algorithm including data preprocessing, feature extraction,and classification. By designing remote controller with communication function, an electroencephalography-based(EEG-based) brain-computer interface(BCI) for television remote control asynchronous system is developed. The system can help the patients suffering from Amyotrophic Lateral Sclerosis(ALS) to turn on, turn off, select TV channels and adjust volume using their brain waves. Five healthy subjects took part in train experiments and online experiments;the results demonstrate that the averaged information transfer rate of the system is approximately 35 bits/min and all the subjects can control the system efficiently with an average accuracy of 87% after a special training procedure.Secondly, An intelligent wheelchair is devised, which is controlled by a coordinated control mechanism based on brain-computer interface(BCI) and speech recognition.By performing relevant activities, users can navigate the wheelchair with four steering behaviors(start, stop, turn right and turn left). Five healthy subjects participated an indoor experiment, the results not only demonstrated that the efficiency of coordinated control mechanism with a satisfied path opt. ratio and time opt. ratio but also shown that speech recognition is a valuable supplement(fast and accurate) as an assistant for BCI-based control system. The proposed intelligent wheelchair is especially suitable for the patients suffering from paralyzed and aphasic who can initiative to pronounce even a signal voice(e.g. “ ah ”).Thirdly, wheelchair control requires multiple degrees of freedom and fast intention detection, which makes electroencephalography(EEG)-based wheelchair control a big challenge. In our previous study, we have achieved direction(turning left and right)and speed(acceleration and deceleration) control of a wheelchair using a hybrid braincomputer interface(BCI) combining motor imagery and P300 potentials. In this paper,we propose hybrid EEG-EOG BCI, which combines motor imagery, P300 potentials, and eye blinking to implement forward, backward, and stop control of a wheelchair. By performing relevant activities, users(e.g., those with amyotrophic lateral sclerosis and locked-in syndrome) can navigate the wheelchair with seven steering behaviors. Experimental results on four healthy subjects not only demonstrate the efficiency and robustness of our brain-controlled wheelchair(BCW) system but also indicate that all the four subjects can control the wheelchair spontaneously and efficiently without any other assistance(e.g., an automatic navigation system).Fourthly, a new method of data analysis for the time-space-frequency motion picture is presented. The time window and frequency bands are trained by train data for different subjects. The CSP algorithm is used to extract the spatial filtering feature and relevance vector machine is used as classifier. By means of III Data IVa competition(imagery motor, training sets small), the average classification accuracy is 90.28%, which provides a new method for the study of the two classification of motor imagery. In addition, for event related potential, the neural network is initialized with the denoise autoencoder,and it is applied to analysis the event related potential. By means of competition III Data set II: BCI(speller paradigm P300) data, all the indexes in exam are satisfactory.The average recognition rates for subject A and subject B are 93.9% and 94.1% respectively by means of random initialization, and which are 95.4%,96.1% respectively by means of DAE initialization.This project is expected to achieve a new approach for real-time control of multidegree-of-freedom systems, which are based on brain-computer interfaces, and provide a new technology to extend and enhance human control ability.