An efficient matheuristic integration with benders decomposition for unmanned aerial vehicle routing problem in forest fire surveillance

Wildfires annually cause extensive and immeasurable harm to the environment, incurring significant costs for containment efforts. Early detection plays a crucial role in preventing the escalation of wildfires, requiring accurate and frequent updates on images and information. Despite advancements in detection methods utilizing image processing and satellite monitoring, the demand for improved precision and frequency persists. Presently, satellite images can detect fires as small as 0.1 ha with an accuracy of 1 km. While helicopters offer detailed information, their use is both hazardous and expensive. There is growing interest in Unmanned Aerial Vehicles (UAVs) for wildfire detection due to their ability to quickly cover potential fire areas using high-definition and thermal infrared cameras, particularly in remote and hazardous terrains. This study addresses the routing and scheduling of UAVs for forest fire surveillance. A mixed-integer linear programming (MILP) model is formulated, followed by the introduction of matheuristic approaches to tackle and solve the problem, drawing inspiration from variable neighborhood search. Matheuristics, which integrate mathematical programming techniques and heuristics, provide efficient optimization methods. Additionally, we incorporate Benders Decomposition to enhance the optimization process by decomposing the problem into a master problem and a subproblem, allowing for more effective exploration of feasible solutions. To evaluate the effectiveness of our proposed solution approach, computational experiments are conducted using a set of instances generated from the Covering Vehicle Routing Problem (CoVRP) and Electric Vehicle Routing Problem (EVRP) datasets. The results indicate that our proposed matheuristics approach, augmented by Benders Decomposition, is capable of producing high-quality solutions. Furthermore, a case study is presented focusing on the Belgrade Forest, one of the largest forests in Istanbul, Turkey, to underscore the benefits of utilizing UAVs in forest fire surveillance.