Fast Multi-command SSVEP Brain Machine Interface without Training

摘要

We propose a new multi-stage procedure for a real time brain machine/computer interface (BMI) based on the Steady State Visual Evoked Potentials (SSVEP) paradigm elicited by means of flickering checkerboards. The developed system work in asynchronous mode and it does not require training phase and its able to detect fast multiple independent visual commands. Up to 8 independent commands, were tested at the presented work and the proposed BMI system could be extended to more independent commands easily. The system has been extensively experimented with 4 young healthy subjects, confirming the high performance of the proposed procedure and its robustness in respect to artifacts.