For several years, a clear trend toward greater automation and reduced crew size has been observed in the shipping industry. The main reasons for this are the reduction of hazards, cost savings, reducing people’s exposure to hazardous situations, and energy and emission reductions. These developments contribute to improved social, environmental, and economic sustainability in maritime shipping. Given the advances in computer technology, artificial intelligence, and their application in other mobility sectors such as aviation and the automotive industry, the implementation of fully autonomous shipping appears to be technologically feasible in principle. However, a prerequisite for autonomous ships to participate in maritime traffic is strict adherence to international safety regulations, particularly the International Regulations for Preventing Collisions at Sea (COLREGs). For autonomous navigation, a ship must precisely perceive its surroundings. This requires the evaluation of data from a multitude of sensors to ensure reliable situational awareness even under challenging conditions, such as bad weather. The processing and fusion of the sensor data must occur autonomously and in real time. In addition to rule-based methods, artificial intelligence techniques, especially deep learning and neural networks, are increasingly being used. A significant drawback of these approaches, however, lies in their lack of transparency, as decisions are often difficult to understand. In 2021, the International Maritime Organization (IMO) investigated the extent to which existing regulations need to be adapted for autonomous ships and defined four levels of autonomy, from automated ships with crews to fully autonomous ships without crews. This research project aims to develop AI methods for generating explainable recommendations for the operation of a ship with Level 1 autonomy. To this end, a hybrid ship handling assistance system will be created that provides transparent and comprehensible recommendations, while the final decision-making responsibility remains with humans. Such a system represents an important step toward autonomous shipping and can contribute to the evaluation and certification of AI-based navigation systems in the future.
Partners
marinom GmbH Jade Hochschule Wilhelmshaven/Oldenburg/Elsfleth
