Optimal Flight Planning for Marine Mammal Surveillance from a Seaplane Unmanned Aerial Vehicle

Unmanned aerial vehicles are playing an increasing role in environmental monitoring and wildlife conservation activities. This paper describes an optimal flight planning algorithm for a solar-powered unmanned aerial vehicle designed to search for and identify whales in the open ocean over long-duration deployments spanning several weeks or months. The algorithm makes decisions about whether the aircraft should take off, cruise, land, or float to save energy at each decision stage, based on battery state of charge, weather conditions, and whale observation probability. This decision-making problem is cast as a Markov decision process, and an optimal controller is derived using Bellman recursion. A simulation environment is constructed using reanalysis weather data and whale observation models derived from empirical data. Simulation results show that the optimal control algorithm significantly outperforms a standard threshold-based decision-making algorithm in terms of increased whale sightings and reduced failure rates. Furthermore, simulation results provide insight into the optimal battery size for the unmanned aerial vehicle, which is seen to exhibit some variation as a function of the latitude and longitude at which the aircraft is deployed.