Serious Sailing: Time-Optimal Control of Sailing Drones in Stochastic, Spatiotemporally Varying Wind Fields

In contrast to traditional mobile robots, renewably powered mobile robotic systems offer the potential for unlimited range at the expense of highly stochastic mobility. Robotic sailboats, termed sailing drones, represent one such example that has received recent attention. After providing a detailed model and corresponding velocity polar for a candidate customized robotic sailboat, this paper presents a stochastic dynamic programming (SDP) approach for time-optimal control of sailing drones in a stochastic wind resource, which provides a feedback control policy to minimize expected time to a prescribed waypoint. The paper provides a Monte Carlo study of the impact of wind direction volatility on the resulting routes, along with an assessment of robustness to mismatches between actual and assumed volatility.