A non Invasive Brain-Computer-Interface for Service Robotics

A Brain-Computer Interface (BCI) enables individuals to control a system solely through their brain activity, without relying on physical movement. These interfaces have numerous applications, particularly in assisting individuals with paralysis. Our research paper details a BCI interface that can classify and control seven wheelchair movements: forward, backward, left, right, stair climbing upwards, stair climbing downwards, and stop. We collected raw signal data using the electroencephalog-raphy (EEG) technique from healthy volunteers, which we then filter before feeding into the feature extraction and classification stages. We evaluated our approach using three classification algorithms: Convolution Neural Network (CNN), Support Vector Machines (SVM), and Random Forest Classifier, and compared their performance. Our experimental results demonstrate that our proposed approach is highly promising for implementing BCI, with a classification accuracy of 99% using a Random Forest Classifier.