Differentiable Route Parametrization and Evolutionary Optimization for Efficient Maritime Routing
Date:
Planning vessel routes over long distances can lead to drastic savings in transport time, fuel consumption, or both. Rafael Ballester tackled this as an optimization problem: we parametrize ship routes using Bézier curves of order n and define a cost function that takes into account time-varying ocean currents. We employ evolutionary algorithms, particularly the Covariance Matrix Adaptation Evolution Strategy (CMA-ES), to efficiently and in parallel find local minima of this functional. During our discussion, we delve into the quality of solutions found based on the vessel’s relative speed in water, the landscape of ocean currents, and the presence of land.
Our research highlights the immense potential of differentiable route parametrization and evolutionary optimization in achieving efficient maritime routing. Join us as we explore the cutting-edge methods that can revolutionize the navigation of vessels, contributing to significant improvements in transportation efficiency and environmental sustainability.
You can watch the talk here: video.
The research for this project is funded by BBVA Foundation and Agencia Estatal de Investigación.