A Multimode Visual-Inertial Navigation Method for Fixed-wing Aircraft Approach and Landing in GPS-denied and Low Visibility Environments

The landing safety is still one of the greatest challenges for fixed-wing aircraft, meanwhile not all airport runways are equipped with instrument landing system (ILS), and global positioning system (GPS) is not always available. It is necessary for aircraft to autonomously acquire high accuracy and stable navigation data during approach and landing in case of GPS loss and ILS unavailability. In this paper, we propose a multimode visual-inertial navigation method that combines the inertial measurement unit (IMU), the forward-looking infrared (FLIR) camera and the barometer into unscented kalman filter (UKF) adaptively according to flight height. In order to obtain accurate motion estimation, different sensing measurements are used at different flight phase. This proposed method run on three significant stages: 1) initial approach (flight height: 1000-200 feet), runway azimuth information and inertia data are fused to estimate aircraft motion; 2) precision approach (flight height: 200-100 feet), sparse runway features extracted from the FLIR images, airport geo-information and inertia data are integrated to obtain precise motion estimation; 3) precision landing (flight height: 100-0 feet), visual simultaneous localization and mapping (VSLAM)and inertial measurements are used to estimate motion states. A general aircraft is elaborately equipped with a short-wave infrared (SWIR) camera, an IMU, a barometer and a flight data recorder to collect real flight data for algorithm verification. Finally, the experimental results demonstrate that the proposed approach can be used for fixed-wing aircraft accurate approach and landing in low visibility and GPS-denied environments robustly. Furthermore, its accuracy has been reached to the level of Inertial/Differential GPS integration.