Optimal and Reinforcement Learning-Based Control of a Motion-Powered Winder for Wave Energy Harvesting

Autonomous sailboats for long-term ocean monitoring face significant energy challenges. While solar panels and indirect wind energy capture (e.g., hydroturbines installed below the hull) are common power sources, this work explores wave energy harvesting as a complementary solution. We investigate the use of an unbalanced mass in the sailboat’s keel to capture wave energy, employing optimal control and reinforcement learning strategies to maximize power generation. Our results demonstrate that an optimal control strategy can enhance power generation efficiency by 70%, while a real-time reinforcement learning-based control algorithm can achieve a 20% improvement under simulated oceanic conditions. The latter algorithm learns and adapts to the dynamics of motion in changing sea states and varying wave spectra, including both frequency and directional components, to optimize system parameters and maximize energy absorption.