A hybrid computer interface (HCI) system, composed of semi-autonomous navigation, decision and interface switching subsystems, is proposed to implement indoor 3-dimensional (3D) space target search for multi-rotor aircraft. The semi-autonomous navigation subsystem is utilized to provide 2-dimensional (2D) feasible directions for the decision subsystem and avoid obstacles semi-automatically in 3D space for multi-rotor aircraft. The decision subsystem applies the joint-regression (JR) model and spectral power methods to extracting the time and frequency domain features from motor imagery (MI) electroencephalogram (EEG) signals, which are collected by the 6 electrodes. Simultaneously, the support vector machine (SVM) is employed to complete the MI feature classification. The interface switching subsystem employs the continuous wavelet transform (CWT) method to identify electrooculography (EOG) features of eyeblink, and switch interfaces between horizontal and vertical MI tasks by analyzing the eyeblink EOG. The actual indoor 3D space target search experiment shows that the presented HCI system has good adaptability and control stability. Compared with similar methods, the semi-autonomous navigation subsystem reduces the control difficulties, and the control precision is increased by about ±10 cm.