Proposal of a Telepresence Robot Using Brain Computer Interface for People with Motor Disabilities

This work proposes a Brain Computer Interface (BCI) using upper limbs motor imagery tasks for people with severe motor disabilities, which can be employed to send commands to a telepresence robot. The BCI classifies four classes of motor imagery tasks from upper limbs. The system is composed of the following blocks: a Common Average Reference (CAR) filter for signal preprocessing, a Statistical Features Extractor of Time Series, and a K-Nearest Neighbors classifier. Eight healthy subjects participated in the experimental protocol, and performed the motor imagery tasks: flexion-extension (F-E) of the right arm, F-E of the left arm, F-E of both arms, and arms relaxed. The system obtained a success rate of 88% with a kappa coefficient of 83% for four classes of motor intentions. The results suggest the feasibility of the BCI proposed to classify motor imagery tasks, which can be used as commands to control the telepresence robot.