Priority-Aware Unmanned Aerial Vehicles Route Planning for Forest Fire Monitoring Using Bayesian-Tuned Hybrid Metaheuristics

Forest and land fires remain a recurring environmental challenge in large forested regions, where spatially dispersed hotspots make timely monitoring and mitigation difficult. Unmanned aerial vehicles (UAVs) provide a flexible platform for rapid aerial surveillance, but effective route planning must account for travel distance, endurance limits, hotspot distribution, and inspection priorities. This study proposes a priority-aware UAV route-planning framework for forest fire monitoring, with experimental evaluation based on hotspot data from Kalimantan, Indonesia. The framework includes hotspot organization, supports short-horizon replanning across two consecutive monitoring cycles, and generates map-based visual outputs to support operational interpretation. To efficiently solve the planning problem, this study adopts an evolutionary computation perspective by developing a Bayesian-optimized GA-VNS method that combines Genetic Algorithm search with Variable Neighborhood Search intensification. Experimental results on the Ketapang and Melawi datasets show that the Bayesian-tuned GA-VNS consistently delivers feasible routing performance, with the largest distance reduction reaching 19.4% over GA in the Melawi Cycle 2 case. Sensitivity analysis further shows that hotspot prioritization improves the inspection order of high-priority locations while preserving strong routing performance. Overall, the findings suggest that the proposed framework provides practical decision-support value for wildfire monitoring and UAV deployment planning.