UAV pose estimation using IR and RGB cameras

In this paper, we propose a system for UAV (Unmanned Aerial Vehicle) 6D pose estimation. The proposed system is experimentally evaluated and verified in a simulated environment. A pair of infra-red active markers and a colored passive marker are utilized to mark the relative pose of the target. Accordingly, two different types of sensors, including the IR (infra-red) and color camera are applied. A PSO (Particle Swarm Optimization) algorithm on the iteration-varying weight is employed to solve the optimal pose of the UAV. The method can be applied to assist the existing IR-marker-based pose estimation system to locate the target when there are not enough IR markers to be detected. In addition, this system is also able to perform as a possible solution for multi-robot localization by identifying the passive markers with different colors on different targets.