The Multi-UAV Collaborative Localization Based on Visual and Inertial Sensors

Precise and robust navigation is a fundamental requirement for unmanned aerial vehicle (UAV) to perform various missions. In this paper, we propose a multi-UAV collaborative localization algorithm based on vector field consensus (VFC). VFC is employed to eliminate the incorrectly matched feature points, which significantly affect the localization accuracy. Moreover, we develop a real multi-UAV system with a centralized terminal to optimize the estimated state of UAVs, using transmission control protocol (TCP) as the communication method to overcome the limitations of ROS. Finally, we demonstrate the efficiency and reliability of our algorithm over ORB-SLAM3 in several experiments on both public and self-recorded datasets.