Optimizing airborne wind energy with reinforcement learning

Airborne wind energy is a lightweight technology that allows power extraction from the wind using airborne devices such as kites and gliders, where the airfoil orientation can be dynamically controlled in order to maximize performance. The dynamical complexity of turbulent aerodynamics makes this optimization problem unapproachable by conventional methods such as classical control theory, which rely on accurate and tractable analytical models of the dynamical system at hand. Here we propose to attack this problem through reinforcement learning, a technique that-by repeated trial-and-error interactions with the environment-learns to associate observations with profitable actions without requiring prior knowledge of the system. We show that in a simulated environment reinforcement learning finds an efficient way to control a kite so that it can tow a vehicle for long distances. The algorithm we use is based on a small set of intuitive observations and its physically transparent interpretation allows to describe the approximately optimal strategy as a simple list of manoeuvring instructions.

Automated summary

Airborne wind energy is a lightweight technology that utilizes airborne devices like kites and gliders to extract power from the wind. Conventional methods are unable to optimize the control of these devices due to the dynamical complexity of turbulent aerodynamics. In this study, we propose to use reinforcement learning to solve this problem, which allows the system to learn an optimal strategy through trial-and-error interactions with the environment. Our simulation results demonstrate that reinforcement learning can effectively control a kite to tow a vehicle for long distances, and the algorithm's physically transparent interpretation allows for a simple list of maneuvering instructions to describe the approximately optimal strategy.