LSTM-based Classification of Multiflicker-SSVEP in Single Channel Dry-EEG for Low-power/High-accuracy Quadcopter-BMI System

Quadcopters, typically known as drones, are being used in an increasing range of scenarios such as unmanned aerial vehicles. The goal of this research is to use electroencephalography (EEG) to establish a method for controlling drones using a brain–machine interface system based on the steady-state visual-evoked potential (SSVEP). To reduce the load on participants during a long-time usage, such a system must be simplified. The proposed method is, therefore, limited to one EEG channel. Drones can exhibit five types of movement: taking off (rising), moving forward, turning right, turning left, and landing. Participants are therefore presented with five multiflickers simultaneously. However, concerns arise over the effect on classification accuracy with using only one channel of the SSVEP. We, therefore, evaluated the classification accuracy using long-short-term memory, which is a method of deep learning that has garnered significant attention. After conducting an experiment with four healthy men, the results indicated a high accuracy of 96% on average. A second experiment was conducted in which the three participants flew actual drones in a series of movements consisting of taking off, moving forward, and landing. We subsequently compared the accuracy of those movements and the flight times.

[1]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[2]  R. Fisher.  THE USE OF MULTIPLE MEASUREMENTS IN TAXONOMIC PROBLEMS , 1936 .

[3]  Minho Kim,et al.  Quadcopter flight control using a low-cost hybrid interface with EEG-based classification and eye tracking , 2014, Comput. Biol. Medicine.

[4]  Corinna Cortes,et al.  Support-Vector Networks , 1995, Machine Learning.

[5]  Meng Wang,et al.  A Wearable SSVEP-Based BCI System for Quadcopter Control Using Head-Mounted Device , 2018, IEEE Access.

[6]  N. Altman.  An Introduction to Kernel and Nearest-Neighbor Nonparametric Regression , 1992 .

[7]  Saeid Nahavandi,et al.  A time domain classification of steady-state visual evoked potentials using deep recurrent-convolutional neural networks , 2018, 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018).

[8]  Michal Jakubczyk,et al.  A framework for sensitivity analysis of decision trees , 2017, Central European Journal of Operations Research.

[9]  John Thomas,et al.  Deep learning-based classification for brain-computer interfaces , 2017, 2017 IEEE International Conference on Systems, Man, and Cybernetics (SMC).

[10]  Toby P. Breckon,et al.  On the Classification of SSVEP-Based Dry-EEG Signals via Convolutional Neural Networks , 2018, 2018 IEEE International Conference on Systems, Man, and Cybernetics (SMC).