Review of scene matching visual navigation for unmanned aerial vehicles

Abstract

High precision location and navigation is one of the key techniques of autonomous flight, efficient reconnaissance, and precision strike of unmanned aerial vehicles (UAVs). Combined with inertial system, scene matching visual navigation technology can be used in a highly precise autonomous navigation system because of its simple structure, passive working mode, and high locating precision. Many previous studies on visual navigation systems of UAVs focus on extracting information of aircraft attitude and navigation using visual information, or the combination of the navigation system and IMU sensors. In China, numerous researches have focused on the technologies of the inertial-combined navigation, 3D map reconstruction with image or laser, and visual navigation landing application. In this study, we summarize the characteristics, categories, and analytical methods of scene matching visual navigation for UAVs. Then, we conclude the primary techniques in high precision, strong real-time, robust and continuous visual navigation tasks under different time, visual angle, illumination, resolution, platform, and sensor.