EEG-Based Motor Imagery Classification and its Implementation

Abstract

This project aims to develop an EEG-controlled robotic arm for individuals with motor impairments, offering them increased autonomy and functionality in their daily lives. The system utilizes a dataset of pre-recorded EEG signals associated with motor imagery and leverages a Convolutional Neural Network (CNN) model to classify incoming EEG signals, enabling users to control the robotic arm through their thoughts. The dataset consists of EEG signals generated by the imagination of performing specific hand and arm movements, allowing users to manipulate the robotic arm's motion and perform various tasks simply by thinking about the movements. The CNN model will be trained and evaluated using the collected EEG dataset, and subsequently deployed on a portable Raspberry Pi 4B platform. This implementation ensures the EEG-controlled robotic arm is lightweight, adaptable, and easily accessible to users. By allowing individuals with motor impairments to control a robotic arm using their brain signals, this project holds tremendous potential to enhance their quality of life, empowering them to perform tasks independently and with greater ease.