Full steam ahead for autonomous shipping

Road freight transport has nearly doubled in the past 30 years, leading to increased congestion and emissions along transport routes. Currently, only around seven percent of goods in Germany are transported via inland waterways. "Shifting some of the transport to rivers and canals is a crucial contribution, not only to alleviating road congestion but also to making transport more environmentally friendly by using new technologies," says Thoralf Noack, Coordinator for Maritime Research at DLR.

A digital test field for automated and autonomous inland navigation is currently under construction on the Spree-Oder Waterway in Germany. In contrast to shipping on the high seas, the reception of satellite data for navigation in built-up areas can be unreliable, particularly around bridges or locks. However, reliable positioning is all the more important in these areas in order to safeguard structures from collisions and prevent traffic disruptions. The solution lies in the intelligent integration of satellite navigation with other sensors installed both on the ship and on land, such as in ports or on locks. These sensors include cameras, lasers or radar sensors that measure environmental data with centimetre precision.

The positioning system relies on global navigation satellite systems like GPS and Galileo to determine its position. Nonetheless, achieving the necessary precision for a vessel to safely enter a lock with centimetre accuracy requires additional correction data. This data will be made available via VDES (VHF Data Exchange System). VDES, the successor to AIS (Automatic Identification System), is a radio system designed to enhance the safety and management of shipping traffic through the exchange of navigation and other ship data. In contrast to AIS, VDES boasts a higher bandwidth, facilitating more interference-free data transmission between ships, base stations and traffic centres. The development of VDES has greatly benefited from the profound expertise of scientists at the DLR Institute of Communications and Navigation, who are actively involved in its ongoing international standardisation process.

eMIR maritime testbed in the North Sea

To test the reliability and effectiveness of automated systems at sea, DLR operates the e-Maritime Integrated Reference Platform (eMIR) testbed together with authorities and industry. Here, highly automated assistance systems are researched and developed, for example for collision avoidance. The testbed comprises two modules: a simulator and a real-world test environment at sea. The virtual infrastructure includes AI-based environmental, traffic and ship simulators. With their help, the risks and the efficiency of innovative maritime systems can be assessed while the product is being developed. The physical testbed – an area of sea from Brunsbüttel past Cuxhaven and Wilhelmshaven to Heligoland equipped with instrumentation for this purpose – allows maritime systems to then be assessed under real-world conditions.

For autonomous vehicles to navigate effectively, they must be able to recognise their surroundings and nearby objects with the utmost precision. Systems and processes employing artificial intelligence (AI) prove well-suited for this purpose. Their programs demand a high level of trustworthiness to ensure they are not prone to errors. Like humans, AI must undergo training to recognise and navigate around obstacles such as ships, buoys, floating containers, people or marine animals. In optical camera-based systems, AI is trained using images of ships from various perspectives. However, there are challenges posed by changing environmental conditions, such as water droplets on the camera lens, which can hinder the accurate identification of objects. Certifying these AI systems presents its own set of challenges, too. To address this, DLR operates a ship bridge simulator at the Federal Maritime and Hydrographic Agency (BSH). Collaboratively, DLR and the BSH have developed a comprehensive test catalogue for this simulator, encompassing thousands of scenarios that the AI must swiftly recognise.

MUM's primary applications are pipeline maintenance and the detection and removal of munitions discarded during the Second World War. "The advantage lies in its ability to autonomously conduct repairs to critical infrastructure such as gas pipelines, throughout the year, even underwater in adverse weather conditions like storms and waves," explains Carl Wrede, Deputy Director of the DLR Institute for the Protection of Maritime Infrastructures. The strategy involves the large mothership positioning itself as close as possible to a site, where it will deploy a smaller submarine to carry out autonomous work using a gripper arm.

Simultaneously, efforts are underway to establish a comprehensive legal framework. The legal landscape on the high seas transcends national boundaries. Different regulations govern navigation depending on the type of vessel – whether it is a motor vessel, sailing boat or fishing boat – even within German territorial waters. Social acceptance, legal considerations and the reliability of AI constitute crucial parameters that influence the technical design of the ship. MUM's initial testing will take place in a maritime testbed in the Baltic Sea, at an average depth of 70 metres.