A Monocular Vision-Based Relative Distance Calculation Method for Fixed-Wing Multi-UAVs

The cooperative operation of multi-UAVs systems in complex environments heavily relies on accurate relative distance perception. However, traditional wireless communication-based distance perception methods exhibit significant limitations in environments with electromagnetic interference or unstable signals. To address the problems, this paper proposes a monocular vision-based method for calculating relative distances in fixed-wing multi-UAVs systems. First, a reference image library containing deep information is constructed for the target UAV from different viewpoints. During the actual flight process, the proposed structure-guided viewpoint selection and orientation alignment method is employed. Additionally, the deep feature matching model established in this paper is used to associate real-time images with the reference image. Then, 2D-3D point correspondences are formed based on the depth information, and finally, the relative distance is calculated by combining the PnP algorithm with the RANSAC method. Simulation experiments were conducted using two fixed-wing UAVs for validation, and the results show that the method achieves a maximum relative error of less than 3.5%, with an average relative error of 0.54%. The results demonstrates that the proposed method provides a robust and accurate monocular vision-based relative distance perception solution for UAV cooperative tasks.