GNSS 불용 구간 내 특수 임무용 드론 자세 보정을 위한 영상 정보 기반 적응형 항법 알고리즘

Abstract

In the field of UAV (Unmanned Aerial Vehicle), drone is noticed that it can replace human role in unknown/hazardous area. Drone’s position is calculated with GNSS (Global Navigation Satellite System) and INS (Inertial Navigation System), however, environment like under bridge, GNSS signal is not reliable, which makes INS diverge. In this situation, drone’s orientation is important for suppressing the divergence of position by INS. Roll and pitch angle can be estimated by employing G-slave mode attitude compensation algorithm. However, yaw angle cannot be estimated well. In this paper, we propose indirect EKF (Extended Kalman Filter) based adaptive navigation algorithm that uses line information from image. In the algorithm, we calculate the relative yaw angle from line information. To get reliable yaw angle, we introduce angle compensation algorithm that corrects the distorted line information by roll and pitch angle. Furthermore, we propose outlier mitigation and adaptation rule based on distance between inlier-outlier features and the number of features for classifying the reliability of the yaw information from the image. Through the simulation, it is confirmed that the proposed algorithm estimates yaw angle sufficiently. After that, we conduct experiment and it is validated that proposed algorithm can reduced heading error with 25.32%.