Estudo e comparação de algoritmos para seleção de nós sobre o Problema de Escalonamento de Drone Paralelos com o Caixeiro Viajante

Abstract

The famous Traveling Salesman Problem (TSP) is a challenge that often arises in large companies that handle deliveries, and a new variable is being added to this issue: drones, small unmanned aerial vehicles. Their use has contributed to a new way of thinking about the classic Traveling Salesman Problem. It is now possible to make deliveries with a delivery person and a drone working together to mutually assist each other. This work will address a variant of the problem, called Parallel Drone Scheduling with the Traveling Salesman Problem (PDSTSP). The goal of this new problem is to integrate the use of a drone into a delivery route that was previously served only by a human delivery person. While the person follows the route designed by the Traveling Salesman, they no longer need to visit certain customers because the drone will cover them. Specifically, this work will explore the question of which pre-selection model better chooses the customers that the drone should visit and return directly to the origin. To perform this comparison, two algorithms will be developed: a greedy algorithm and a random algorithm. Python libraries will be used to solve the TSP. Finally, we will obtain a value corresponding to the time it takes for the last vehicle to return to the origin after completing its tasks. The results indicated that the greedy algorithm performed better in the pre-selection of nodes.