Vision-based Navigation for Autonomous Landing System

摘要

In this paper, autonomous landing system for unmanned air vehicles (UAVs) based on visual navigation is presented. Two main problems are given with proposed solutions, including autonomous landing system architecture based on Robot Operating System (ROS) and vision algorithm for realtime landing platform detection and pose estimation. Our system was fully implemented on on-board computer and monocular camera equipped on UAVs. Gazebo Simulator with Iris quad-copter model was used for rapidly testing about vision-based and control algorithm. Also, a series of autonomous flight tests are successfully performed for common scenarios with static and moving targets. Obtained results will be verified via normal flight sensors for guaranteeing qualities of our autonomous landing system.