Collaborative Object Detection and Distributed Fusion Based on Dual Camera

This article investigates the collaborative localization and distributed fusion of multiple unmanned aerial vehicles based on cameras, including object detection algorithms, state estimation based on unscented Kalman filters(UKF), and data fusion. Starting from introducing the model we built in Gazebo and the 3D target localization implemented in real-time operating system(ROS), we analyzed the characteristics of the target detection results and the necessity for improvement. In order to obtain more accurate target positions, we proposed a collaborative localization algorithm based on post-fusion, which was processed and validated in MATLAB. Finally, the effectiveness of our algorithm was demonstrated through comparison.