| Summary: | The Industry 4.0 revolution requires the automation of more industrial processes. Particularly in warehouse logistics, Unmanned Aerial Vehicles (UAVs) are a key technology for the digital transformation of warehouse management tasks, providing the ability to perform automatic real-time inventory counting, localize hard-to-find items, and reach narrow storage areas. The use of this type of robot poses new challenges, such as UAV indoor localization, navigation, collision avoidance, and fleet management. This thesis proposes an indoor navigation system for a small commercial UAV, with a visual-inertial Graph-SLAM approach, and a simple way for executing user-defined behavior tree missions through the Aerostack framework, in order to achieve an easy, fast, and cost-efficient implementation. A system architecture was presented, with the development of Robot Operating System (ROS) interfaces for the multiple hardware and software elements, and the implementation of new behaviors related to inventory management tasks. The system was tested in an indoor environment, where the executed mission allowed the UAV to take off, navigate to two marked locations, take photos of the markers, and return to the take-off location. The system performance was evaluated by comparing sensor data and the vehicle behavior during the mission with the expected behaviors, according to the mission plan.
|
|---|