Electroencephalography-Based Brain-Computer Interfaces for Robots Control Using Deep Learning

摘要

In recent years, there has been considerable progress in the invention of brain-controlled mobile robots and robotic arms. The development of electroencephalography (EEG) technology has made it possible to operate external equipment directly from the brain. A brain-computer interface (BCI) enables the communication between the brain and an external device. It is used in many applications such as security, education, neuro-marketing, entertainment, and medical applications. In this paper, we present an approach to supply the control of a robot based on the identification of given commands. We control the direction of the robot according to the attention level. We also use eye blinking as an alternative way to control the robot’s direction in the left and right directions. We implemented the full pipeline to achieve this goal and verify the correctness of the desired behavior. We start by acquiring the signal, processing, classification, and action execution. For the signal classification step, we apply a deep learning 1D convolutional neural network, which outperformed several classical machine-learning models. The outcomes show that the 1D convolutional neural network is the most proper deep learning method to detect the level of attention with a testing accuracy of 95% to control our robot’s motor direction. In this research, a 1D-convolutional neural network technique based on a deep learning model for attention level, eye blinking to classify the brain wave, and the method to control the robot will be presented.