In recent years, the demand for unmanned aerial vehicles (UAVs) has skyrocketed in various fields such as logistics, real-time monitoring, search & rescue, and communications. Due to the low production and maintenance costs and high mobility, UAVs are expected to be utilized in miscellaneous applications. The biggest hurdle to grafting UAVs in other areas is that expertise in designing UAV architectures and professional maneuvering techniques is required to operate UAVs stably. Therefore, it is quite time-consuming and demanding to develop UAV applications without any help of UAV hardware architecture experts and experienced pilots. To simplify the design process of drone architecture, this thesis presents an algorithm for selecting UAV hardware components. The test flight showed the stable maneuvering of the drone assembled with the hardware selected by the proposed algorithm. Besides, this thesis proposes a UAV simulation platform capable of designing a system for various missions by modifying only modularized custom nodes without prior knowledge of UAVs. The simulation platform is an integrated system consisting of Software-in-the-Loop (SITL), Robot-Operating System (ROS), off-board computer, and physics engine named Gazebo. This paper introduces two application instances as simulation platforms. One instance is a drone base station simulation combined with a channel simulator, and the other is reinforcement learning (RL)-based simulation.
Publisher
Ulsan National Institute of Science and Technology (UNIST)