Atmospheric Energy Harvesting for Small Uninhabited Aircraft by Gust Soaring

Abstract

Applying bird-inspired flight planning and control techniques to small robotic aircraft can greatly improve flight performance. This paper discusses a method for improving cruise performance of an uninhabited glider by harvesting energy from short period stochastic phenomena (i.e. gusts). Receding horizon control is used to plan a sequence of control inputs that maximizes an energy- based reward function over a time horizon, using only local knowledge of atmospheric conditions. Parameters of the receding horizon controller and parameters in the reward function are tuned using an evolutionary algorithm. The resulting controller is tested using Monte Carlo simulations of flight through Dryden gust fields: results show significant improvement over constant speed flight. Robustness of the receding horizon control approach to changes in aircraft parameters (parasite drag) is also evaluated using Monte Carlo simulations: increasing parasite drag results in gracefully degrading performance over the nominal condition while decreasing parasite drag results in increasing performance.