An autonomous UAV-based monitoring device for precision viticulture

Current advancements in proximal sensing for sustainable viticulture focus on in-tegrating autonomous systems, such as UAVs, with miniaturized sensors and machine learning to enhance the efficiency of vineyard monitoring. The study, part of the COLIBRI' project (Collaborative integration of mini-UAV, miniatur-ized sensors, and machine learning for proximal sensing towards sustainable viti-culture), developing an autonomous drone system equipped with advanced sen-sors to automate vine monitoring. The drone carries a thermal camera to assess leaf water stress and a hyperspectral camera to evaluate grape ripening. The on-board integration of a companion computer (Raspberry Pi) enables real-time data analysis using machine learning models. This approach facilitates in-stant classification of ripening conditions and water stress levels, supporting vineyard management decisions. The UAV autonomously collects data from specified targets, processes inputs ex-ternally, and executes assigned missions reliably. It is equipped with a flight con-troller to stabilise flight and control engine functions and a companion computer to manage software and instruct the flight controller. A ground station monitors the mission via telemetry and uploads data to the UAV. The system uses a GPS receiver for navigation, while the object approach relies on artificial vision with Aruco fiducial markers and the OpenCV2 library. Test flights demonstrated reli-able mission execution with potential for the future.