Flight plan optimization for multiple drones in construction sites

• Autores: Sotelo Vila, Alvaro, Ramírez Cerna, Lourdes
• Formato: artículo
• Fecha de Publicación: 2023
• Institución: Universidad de Lima
• Repositorio: Revistas - Universidad de Lima
• Lenguaje: español
• OAI Identifier: oai:revistas.ulima.edu.pe:article/6230
• Enlace del recurso: https://revistas.ulima.edu.pe/index.php/Interfases/article/view/6230
• Nivel de acceso: acceso abierto
• Materia: dynamic programming; flight planner; surveillance drones; genetic alorithm; programación dinámica; planificador de vuelo; drones de vigilancia; algoritmo genético

The construction sector is searching for new technologies, such as drones, that prove helpful for the surveillance and supervision of construction sites, especially in pandemic situations. This research proposes designing flight planning models to optimize flight time and speed. The main objective is to develop a model that allows multiple drones to carry out supervision tasks in construction areas. In this regard, it presents a dynamic programming model and a metaheuristic based on genetic algorithms, both applied for optimizing flight plans with multiple drones. The development process involves planning the route that the drones will follow by formulating the problem with the corresponding parameters. The next step is to generate a model for the dynamic programming algorithm, which is then validated using a genetic algorithm. The proposals implemented in Python were tested in 14 scenarios, gradually increasing in complexity. The dynamic programming-based model significantly improves planning time in all scenarios, achieving an average difference of 281,34 seconds or 4 minutes and 47 seconds, 98,01 % better than the genetic algorithm. Additionally, there is a considerable improvement in segment speeds, as the results show. A paired test evaluated these advancements. The hypothesis is supported with a p-value of 0,0031 for time and 0,0071 for the gain obtained by the objective function in both cases. This confirms the superiority of the dynamic programming algorithm compared to the genetic algorithm.