Research on Orthogonal Iteration Algorithm of Visual Pose Estimation for UAV Landing

Aimed at the problem that the performance of orthogonal iterative (OI) algorithm is affected by the initial value when applying it for UAV pose estimation, the solutions of different non-iterative algorithms are selected as the initial values. Algorithm performances are evaluated from three aspects in algorithm stability, accuracy and real-time performance. Computer and experimental simulation are built to verify that a better performance can be obtained by using a Robust O(n) solution to the Perspective-n-Point problem (RPnP) as the initial value. Furthermore, the main factors affecting accuracy are explored among the number of feature points, the distance between the camera and the cooperation target and the random noise.