The advance of MEMS-based inertial sensors successfully expands their applications to small unmanned aerial vehicles (UAV), thus resulting in the challenge of reliable and accurate in-flight alignment for airborne MEMS-based inertial navigation system (INS). In order to strengthen the rapid response capability for UAVs, this paper proposes a robust in-flight alignment scheme for airborne MEMS-INS aided by global navigation satellite system (GNSS). Aggravated by noisy MEMS sensors and complicated flight dynamics, a rotation-vector-based attitude determination method is devised to tackle the in-flight coarse alignment problem, and the technique of innovation-based robust Kalman filtering is used to handle the adverse impacts of measurement outliers in GNSS solutions. The results of flight test have indicated that the proposed alignment approach can accomplish accurate and reliable in-flight alignment in cases of measurement outliers, which has a significant performance improvement compared with its traditional counterparts.
Understanding the factors that influence the quality of unmanned aerial vehicle (UAV)-based products is a scientifically ongoing and relevant topic. Our research focused on the impact of the interior orientation parameters (IOPs) on the positional accuracy of points in a calibration field, identified and measured in an orthophoto and a point cloud. We established a calibration field consisting of 20 materialized points and 10 detailed points measured with high accuracy. Surveying missions with a fixed-wing UAV were carried out in three series. Several image blocks that differed in flight direction (along, across), flight altitude (70 m, 120 m), and IOPs (known or unknown values in the image-block adjustment) were composed. The analysis of the various scenarios indicated that fixed IOPs, computed from a good geometric composition, can especially improve vertical accuracy in comparison with self-calibration; an image block composed from two perpendicular flight directions can yield better results than an image block composed from a single flight direction.