Turn Decisions for Autonomous Thermalling of Unmanned Aerial Gliders

Unmanned Aerial Gliders form a subclass of fixed-wing Unmanned Aerial Vehicles which promise to offer sustained flight for a wide range of applications. Autonomous soaring allows these aircraft to detect and exploit rising air masses (thermals) without user input, which greatly simplifies their operation. While previous research has focused on the detection and exploitation of thermal updraft, the initial turn at the entry point into the thermal has been ignored. This paper explores the initial turn decision at the instant of thermal detection in order to improve soaring performance by flying directly into the thermal. A high-fidelity simulation of an off-the-shelf RC glider is constructed, along with the Weather Research and Forecasting model to capture realistic thermal convection. The effects of turn decisions on thermalling performance are examined through a large batch simulation on a Matlab/Simulink environment. Thermalling algorithms are subsequently integrated into the PX4 flight stack for Software-in-the-Loop simulations and flight tests using a Pixhawk flight controller. Simulated and experimental results demonstrate the importance of turn decisions for improved overall soaring performance.