Convolutional Neural Network Based Electroencephalogram Controlled Robotic Arm

Abstract

In this paper, we present a six-degree of freedom (DOF) robotic arm that can be directly controlled by brainwaves, also known as electroencephalogram (EEG) signals. The EEG signals are acquired using an open-source device known as OpenBCI Ultracortex Mark IV Headset. In this research, inverse kinematics is implemented to simplify the controlling method of the robotic into 8 commands for the end-effector: forward, backward, upward, downward, left, right, open and close. A deep learning method namely convolutional neural network (CNN) which constructed using Python programming language is used to classify the EEG signals into 8 mental commands. The recall rate and precision of the 8 mental command classification using the CNN model in this research are up to 91.9% and 92%. The average inference time for the system is 1.5 seconds. Hence, this research offers a breakthrough technology that allows disabled persons for example paralyzed patients and upper limbs amputees to control a robotic arm to handle their daily life tasks.