Methodology for operating agricultural UAVs with automatic calibration and semantic layout of flight missions

Abstract

This paper presents a functional model of a group of unmanned aerial vehicles (UAVs) for liquid fertilizer application, as well as a method for automating the preparation and execution of flight missions formulated as a skill-based system. The first part of the study identifies the primary challenge as high cognitive workload. A model of liquid substance application is then introduced, describing a closed-loop process of planning – task allocation – synchronized execution – monitoring, considering UAV technical constraints and agronomic requirements. The theoretical foundations for mission parameter calculation are examined in detail, including application rate of the working solution, pump performance, flight speed, swath width, and pass overlap percentage. An automatic calibration algorithm for the liquid application system is proposed, in which the ground control station compares theoretical and actual flow parameters and provides a ready-to-use solution. A method for decomposing UAV missions into atomic tasks is proposed, distinguished by their automatic composition into a unified skill based on the semantics of the user request. This approach enables transferring analysis and decision-making functions from the human operator to a computational system. The practical significance of the work lies in simplifying mission preparation, minimizing human-factor errors, and improving the efficiency of cooperative UAV deployment in agricultural applications.