Descrição
The United Nations (UN) estimates that the population and food demand will grow significantly until 2050. Researchers have been working to improve the food production chain, and automation and robotics are great allies. In the technological and agricultural field, precision farming, through the decision support system, improves farming management and optimizes production while preserving resources. In crops, this process relies on measures conducted throughout the cultivation period, which allows punctual and assertive actions. The crop monitoring usually is performed by a Unmanned Aerial Vehicle (UAV), which involves the coverage of an Area of Interest (AOI). The solution to the Coverage Path Planning (CPP) problem defines the sequence of waypoints the UAV must visit to cover the entire area. This problem is classified as NP-Hard. Unfortunately, because of its limitations, a single UAV can not perform a coverage task in large areas without recharging or changing the battery. In this case, multiple UAVs can cover the area. However, it increases the complexity of the problem, once it is necessary to distribute the AOI among the UAVs in a balanced way while optimizing the coverage path of each one. This work conducted a Systematic Review (SR) showing the main methods used to address the Multiple Coverage Path Planning (MCPP). These methods include algorithms created expressly for this topic, clustering techniques, sophisticated Mixed Integer Linear Programming (MILP) models solved by mathematical solvers, and Artificial Intelligence (AI) solutions. The SR clarified that there are neither studies investigating the wind effects in the MCPP, nor considering a mobile battery recharge station, and the multi-agent systems approaches are in early development phases being tested only in simulation. Therefore, it is possible to contribute to the field by considering the wind in the MCPP in a multi-agent architecture where a land-based Mobile Robot (MR) can serve as a recharge station. The thesis proposal is organized into four major steps: (i) creating an environment model to consider the influence of the wind in the MCPP; (ii) developing a MILP model to compute optimized solutions on small-scale problems; (iii) building a multi-agent architecture to handle the MCPP in medium and large scale; (iv) adapting the multi-agent solution to handle a mobile battery recharge station.Quando?
