A Brain to UAV Communication Model using Stacked Ensemble CSP algorithm based on Motor Imagery EEG signal

The creation of non-invasive Neuro-headset technology has resulted in a variety of applications such as control, automation, and vehicular control. Unmanned Aerial Vehicles (UAVs), mostly used in aerial communication, is a growing field of study that may be applied to both defense and commercial applications. As the need for drone control grows, recent advancements in the BCI-based drone control system have been made. This paper proposes a unique prototype to achieve smooth and stable control of a UAV by the human brain based on a motor imagery signal for semi-autonomous navigation. In such cases the accuracy and signal detection and response time get the utmost priority to get a decent performance of the vehicle. Currently, it is very common to use CSP for the classification of motor imagery data. However, this algorithm needs improvement in the low quantity of training samples and noisy data. Because in that scenario the over-fitting may be a problem. To overcome those drawbacks this study proposed Stacked Ensemble Common Spectral Pattern (SECSP). Comparing this method with Common spatial pattern (CSP), Filter Bank CSP (FBCSP), and Common Spatio-Spectral Patterns (CSSP) it has been proved that our model outperforms those popular methods in terms of accuracy, sensitivity, robustness, and precision for motor imagery Electroencephalogram data. The publicly available data-set for motor imagery BCI Competition IV, Data-set 1, BCI Competition III, Data-set IVA have all been tested with Stacked Ensemble CSP (SECSP). Performance is enhanced when compared to previous approaches, with an average accuracy of 87.74 percent for all subjects and 95.64 percent for the first and second data sets, respectively.