Detección y tratamiento de las señales EEG que permitan definir tareas de un drone.

Abstract

This document shows the process to control a drone with EEG signals, using an array of dry electrodes around the motor cortex and in the forehead area, following the 10/20 distribution and filtering the signals in the alpha rhythm. A database is set up with 15 people, of different genders and ages, to train three different computational intelligence paradigms (k-nearest neighbors, random forests, and convolutional neural networks) with the inputs passed through different preprocessing after filtering (without preprocessing, normalization, standardization, scaling between fixed values) with the help of python "scikit" learn. A GUI is made to help control the drone using "pyParrot" in python.It is obtained that the best classification is with the convolutional neural network with the inputs are preprocessing with which you can give orders to the drone in real time.