Indoor space target searching based on EEG and EOG for UAV

This paper puts forward a noninvasive electrooculography (EOG) and electroencephalogram (EEG)-based hybrid computer interface (HCI) system to implement the indoor target searching in three-dimensional (3D) space for a low-speed multi-rotor aircraft. The HCI system mainly consists of three subsystems, including the interface switching, decision and semi-autonomous navigation. The interface switching subsystem is accomplished by detecting the eyeblink EOG. The continuous wavelet transform is employed to indentify eyeblink features which are used to switch interfaces between horizontal and vertical motor imagery (MI) tasks. The average accuracy of the eyeblink feature detection reaches to 97.95%. The decision subsystem employs the joint regression (JR) model and spectral powers methods to extract the time and frequency domain features of MI tasks by analyzing the left- and right-hand MI EEG. Simultaneously, the support vector machine is applied to accomplish the MI tasks classification and final decision. The average classification accuracy of the HCI system reaches to 93.99%. The semi-autonomous navigation subsystem extracts the environmental features to avoid obstacles semi-automatically in 3D space and provide feasible directions for the decision subsystem. The actual indoor 3D space target searching experiments are put forward to verify the feasibility and adaptation performances of this proposed HCI system.