A Novel Event-Related Potential-Based Brain–Computer Interface for Continuously Controlling Dynamic Systems

Abstract

Using electroencephalograms (EEG) to continuously control dynamic systems (such as wheelchairs and vehicles) have important values in helping people improve their quality of life and independence. In this paper, we develop a novel event-related potential (ERP)-based brain-computer interface (BCI) to detect human intention for continuously controlling dynamic systems. First, channel selection is performed by using the improved forward floating search algorithm (ISFFS). Then, a convolutional neural network (CNN) is used for feature extraction and a linear discrimination analysis or support vector machine classifier is used to decode intention for each subject. The experimental results of eight subjects show that the proposed system based on the optimal channels selected by the ISFFS algorithm and features extracted by the CNN performs well. Compared to other similar BCI systems, the proposed one can issue a command more quickly and accurately. This work not only facilitates the research of brain-controlled dynamic systems, but it also provides some new insights into the research of BCIs.