Vision-based UAV autonomous landing system

Abstract

Unmanned aerial vehicles (UAVs) have been widely used in suburban traffic planning and guidance, but their short endurance and limited range still affect the application prospect of UAVs. Therefore, the heterogeneous system between UAVs and intelligent vehicles (IVs) can effectively solve this problem. We propose a vision based UAV autonomous landing system and construct a traffic simulation scene based on a complex urban road. This paper uses the gazebo-based simulation platform to test the whole heterogeneous autonomous landing system in different conditions. The quadrotor UAVs are equipped with a monocular camera to detect landmarks. Then, the PID-based controller is used to autonomously land our UAVs. The experiments are conducted using different flight heights. The experimental results indicate that:1) the simulation platform can quantitatively analyze heterogeneous autonomous landing system performance. 2) sufficient adjustment time and higher landmark recognition metrics can effectively improve the landing performance of UAVs.