Planejamento de Rotas de Cobertura para Veículos Aéreos Não Tripulados Aplicados no Controle Biológico de Pragas em Cenários Reais de Grandes Áreas

Abstract

Biological Control (CB) is a sustainable method for managing agricultural pests, us ing natural agents to regulate their populations. The use of Unmanned Aerial Vehicles (UAVs) to release these biological agents offers advantages such as low operational costs and high efficiency over large areas. However, the current Coverage Path Plan ning (CPP) methods for UAVs in CB are empirical and heavily reliant on the pilots’ experience, lacking a systematic approach that considers all critical variables. Further more, the existing literature does not fully address all aspects relevant to CB. In this context, the objective of this thesis was to enhance the generation of trajectories for UAV operations in the release of biological agents across large areas, utilizing CPP concepts. To achieve this goal, a framework and two algorithms were developed and validated. The framework was implemented as a web application that automates the generation of routes, incorporating the proposed algorithms. The first algorithm, "Sub routes," decomposes the complete route into sub-routes, applying a back-and-forth pattern to minimize path overlap and optimize flight time by aligning the trajectory with the largest edge of the area to be covered by the UAV. The second algorithm, "Sub areas," also applies the back-and-forth pattern but proposes a decomposition of the area of interest into subareas using the Voronoi diagram, aiming for a more balanced division, especially in areas with concave internal angles. The validation of these al gorithms was conducted through experiments involving real flights in three different scenarios with various payload configurations. Given that a biological agent release device was attached to the aircraft, the experiment also evaluated the potential influ ence of the payload on coverage. The quantitative results show that, when using the Sub-routes Algorithm, the total area covered was 114.61 hectares, with a total distance of 29.7 km covered in 44 minutes and 48 seconds. In contrast, the Subareas Algorithm resulted in a slightly shorter distance of 28.9 km covered in 42 minutes and 57 sec onds, representing a reduction of 2.6 km (approximately 2.8%) in the total distance and 1 minute and 51 seconds (approximately 4.1%) in flight time. Furthermore, the distribution of work among the flights was more balanced with the Subareas Algorithm, where the flights covered 36.5%, 32.5%, and 31.0% of the total area, respectively, suggesting increased operational efficiency. Although both algorithms demonstrated effectiveness, the Subareas Algorithm proved particularly advantageous in scenarios involving areas with concave internal angles, where coverage uniformity and efficiency are crucial. Additionally, the influence of the payload was minimal, indicating that the additional weight did not significantly compromise coverage performance, confirming the effectiveness of the proposed route planning solutions.