Marine guidance using AI: Development of a optimization system for marine vessels

Abstract

Navigating and handling a vessel optimally is by no means an easy task. Not only does one have to find the appropriate path to a destination while avoiding obstacles but also consider external forces such as weather conditions. Further, traveling by water is often very energy-consuming. A guidance system to facilitate such tasks should have great potential, not the least in the form of environmental effects. This thesis has constructed a marine guidance system that utilizes modern optimization methods to enable a more efficient way of voyage planning and handling of a vessel concerning fuel economy. Stochastic optimization and black-box modeling have enabled a system to handle and optimize complex environments using minimal information. The results show that it is possible to create accurate fuel consumption models regardless of the configuration of vessel model and engines. Further, a robust path planning system has been derived that manages to optimize several objectives and help drivers achieve more efficient voyage planning. The system has been shown to outscore the voyage planning capabilities of everyday drivers of the sea. These results prove the capabilities of a field that still has much to explore.