Robot Navigation System Based On Asynchronous Online BCI

The rules of BCI System Based on EEG are from offline to online, from synchronous to asynchronous, and its key is to make the BCI system from the laboratory to practical application. Because of the bottleneck problem of brain mechanism research has not been broken, the researchers still need to solve the problem of real time autonomous control and get rid of the interference between the consciousness and the external devices through a variety of technical means. The asynchronous online BCI technology based on the MI is the most close to practical application.The goal of this thesis is to implement asynchronous online control of the robot, designing experiments for real scene. Study on the advantages and disadvantages of different feature extraction and classification recognition algorithm combination for classification. The research is aimed at the related problems of online feature extraction and classification accuracy of EEG signal. The main research content includes the following three aspects:1) The research of detection algorithm for idle state and task state and the judgment of the subjects’idle state and task state judgment are the key technology of asynchronous brain-machine interface. They are also fundamental difference synchronous online BCI system and asynchronous online BCI system. This thesis has designed two kinds of handling strategies for idle state. Respectively for different strategies different algorithms are used to analyze EEG data of working state and idle state. The better handling strategy of idle state eventually selects through the experiment.2) The optimization study of online BCI system channel. The goal of the research on the online BCI system is to use fewer channels to achieve optimal online control. In this thesis, the ant colony algorithm is adopted to channel for further optimization and the channel which has small effect on the result is reduced.3) Implementation of asynchronous online BCI navigation system. For the content of this thesis, built an online BCI platform of asynchronous motor imagery. The platform includes three modules: offline training which is used for training online recognition model, online simulation test which is used for testing the online accuracy of experimenters and online control which is used for control online. Finally, the robot’s navigation control is realized through the online platform.