Motor Imagery EEG-EOG Signals Based Brain Machine Interface (BMI) for a Mobile Robotic Assistant (MRA)

Abstract

Assistive devices for disabled individuals provide the support to fulfil their activities of daily living. The proposed mobile robotic assistant (MRA) in this paper is capable of providing both mobile and manipulation support for users. The MRA, which consists of an electric wheelchair and a custom developed 5DOF robotic manipulator is controlled by a Motor Imagery (MI) Electroencephalography (EEG) and Electrooculography (EOG) based Brain Machine Interface (BMI) which is proposed in this paper. A custom developed Graphical User Interface (GUI) is utilized to interact with the users and users are able to control either the wheelchair or the robotic manipulator based on a combination of left and right hand MI-EEG signals and EOG signals via this GUI. A Multilayer Perceptron (MLP) Neural Network based classifier is developed to classify the EEG signals of left vs right vs rest. EOG signals (eye-blinks) are used to activate the task on the GUI menu. A set of experiments have been carried out with healthy subjects and the results show the effectiveness of the proposed methods.