Brain robot interface using artificial neural network

Abstract

Recent researches in Brain Computer Interface (BCI) that can decode brain EEG signals has aided in an effective robot control which has led to the raise of Brain Robot Interface (BRI). This project focuses on the accurate classification of the user's Action/Cognitive thoughts, where successful decoding of EEG signals can provide a higher degree of freedom control in BRI applications. The EEG signals from the user's scalp are recorded through a non-invasive electrode and prepossessed to produce a noise free EEG signals. Time-Frequency Analysis techniques are used to extract featured from the EEG signal. In this work an Artificial Neural Network (ANN) machine learning algorithm is used as classifier to learn the EEG signal features for effective output classification. This work presents a performance analysis on the accuracy of the system for the proposed combination of Time-Frequency analysis and ANN algorithm for the EEG feature extraction and classifier respectively.