Artificial Intelligence Promotes Automation On Board And In Ports
Innovation leap watchfree bridge
The major challenges facing maritime transport include coping with the growing volume of trade, improving maritime safety, economic efficiency and environmental friendliness.
In the course of advances in information technology, these challenges have led to the rapid development of autonomous technologies. Within the framework of the BMWifunded research project B ZERO, the Fraunhofer CML is now developing a sensor and navigation system in cooperation with Wärtsilä SAM, Hoppe Bordmesstechnik, NautilusLog, the Bernhard Schulte Group, the Federal Maritime and Hydrographic Agency and the Fraunhofer FKIE. The system should be able to guide a ship autonomously between defined departure and arrival points, so that manning the bridge around the clock is not necessary.
The Fraunhofer CML will develop an artificial intelligence for autonomous navigation by using reinforcement learning in B ZERO. With reinforcement learning a system can train meaningful de- cision guidelines without prior knowledge, only by results or responses to its actions. Reinforcement Learning is already used at CML in the fields of object recognition and robotics, and supports the anticipatory avoidance of collisions and grounding in nautical situations. The AI, which will later take over autonomous navigation in B ZERO, is trained at the CML by simulating nautical scenarios with different parameters such as number of approaching ships, sea area, visibility and weather conditions. The decision component to be trained, e.g. collision avoidance, knows the required state of the- se given conditions and reacts with the learned, appropriate voyage and/ or course changes to ensure a safe passage on a route. The expected result is a prototype system, which will be further developed in the simulation laboratory environment of the CML and validated by future tests on board a cargo ship.
Efficiency boost in image recognition
Great potential for maritime logistics results from the use of AI- supported image recognition, or computer vision in short. In addition to the acquisition of digital images, it enables their processing into highly compressed numerical information that can be further processed by machines. Computer vision is thus a key technology for the automated observation of conditions and the detection of changes. These capabilities enable a wide range of applications in the maritime sector. In maritime shipping, for example, many autonomous manoeuvres depend on the permanent, simultaneous and reliable situational awareness that computer vision enables. Gradual changes, such as erosion of quay walls or deformations of a ship‘s hull, can be detected by computer vision, as can the position of cargo units on board or at the terminal.
The CML supports companies in the maritime industry in identifying and exploring the individual possibilities of computer vision. As part of the COOKIE project, which is funded by the IHATEC programme, a visual damage recognition and image-based repair prognosis of empty containers is being developed using artificial intelligence. This will not only ensure compliance with applicable security standards, but also make inspection procedures at the terminal gate more efficient.
In addition to computer vision, the CML has a broad spectrum of expertise in the field of machine learning and offers comprehensive solutions for AIsupported forecasting and assistance systems, from proof of concept to implementation.
Source: CML Fraunhofer