Vision-based real-time target localization for single-antenna GPS-guided UAV

Abstract

This paper discusses a real-time method to calculate the three-dimensional location of a fixed target detected by a gimbaled camera in a fixed-wing experimental unmanned aerial vehicle (UAV) equipped with a single-antenna GPS receiver on board. The camera was operated by an algorithm that automatically tracked the detected target, and the location was obtained using triangulation. The algorithm, however, suffers from a bias in target localization, because the aircraft Euler angles are not measured directly, but rather inferred from GPS velocity. To enhance the algorithm, the influence of the wind was taken into consideration to improve the localization accuracy, which cannot be deduced using a single-antenna GPS. Using a special maneuver, the circular level flight, wind velocity was estimated by the periodic variation of the aircraft's velocity and angle of attack (AOA) using the camera angles. Then, the pseudopitch angle was compensated for using the estimated results. The experimental data show that applying the pitch angle calibration successfully eliminated almost 90% of the horizontal localization error and the final 3D accuracy was less than 10 m.