About autonomous diving robots, underwater imaging and maritime situational awareness

The different missions of our ROVs and autonomous vehicles:

• Monitor water quality, take water samples, inspect underwater structures or support fish farming. The agile, remote-controlled underwater vehicles of the TORTUGA ROV family, which unlike conventional ROVs are equipped with an internal energy storage system, are used for this purpose.
• Exploring the seabed with high-resolution imaging sensors and underwater drones. In the process, Fraunhofer IOSB-AST's DEDAVE vehicle was able to discover historical finds from the Bronze Age and a flight model of the Canadian jet fighter CF-105 Avro Arrow, which was thought to be lost, at the bottom of Lake Ontario. Fraunhofer IOSB's underwater drone technology also held its own in the development of a lightweight technology for mapping the deep sea: our ARGGONAUTS team finished among the world's top five teams in the Shell Ocean Discovery XPRIZE technology competition. 
• Saving people from drowning. Real-life experiments by Fraunhofer IOSB-AST to develop a water rescue robot (television broadcast in German) are already underway in swimming pools, and rescue in inland waters is being planned.
• Precisely measure waterways above and below the surface and thus support the Digitalization of water infrastructure. In a pre-research project, we are developing a autonomous surface vehicle for underwater and overwater mapping – a catamaran equipped with sensors that automates the navigation and mapping of water surfaces or waterways. In the process, potential obstacles are detected and avoided.

Inspection: Our autonomous underwater vehicles are helpful for exploration and inspection tasks. For example, various vehicles in terms of equipment, size and payload are used in the maintenance of offshore wind turbines under water. Oil and gas pipelines, submarine cables between continents, and sheet pile walls in ports and inland waterways can also be maintained with our practical autonomous diving systems.

Vision: We use new variable image acquisition and processing techniques to increase image quality and visibility underwater. This enables visual-optical inspection of underwater infrastructure even in turbid waters. In addition, a carpet of images can be assembled from many individual images or videos to form a panoramic image - without unnatural image distortions. These overview maps are used to inspect underwater infrastructure (e.g., bridge foundations), document, archive, and detect changes from older overviews.

Our AI-based classification and detection algorithms for image datasets can help detect creatures or objects of interest on the seafloor(German text). For this, we trained an algorithm that uses artificial neural networks to search for and detect targeted structures on the seafloor from large image datasets.

Communication: Can submarines communicate via laser light, and what impact does water turbulence have on this type of communication? To investigate these questions within the framework of our long-standing expertise in adaptive optics, since 2020 we have a water turbulence laboratory that is unique in Europe (see interview in German on page 3 in InfOSB 1/2020 [pdf]). Special mirror devices in the 500-liter water tank stretch the length of the light beam many times over.

We not only develop underwater vehicles for diverse missions, but also offer assistance systems for decision support. Especially with the advent of high data volumes, AI-based procedures support situation awareness and anomaly detection in the maritime context.

In project MARISA, for example, a maritime surveillance system with automatic procedures was developed to ensure applicable safety requirements and environmental protection. In doing so, human decision makers are supported in focusing attention on relevant information.

The XAI toolbox can be used to quickly evaluate various XAI methods for AI procedures. Among others, surveillance bodies of maritime areas or ports can apply this when it comes to identifying ship types based on trajectories.

In the European defense research project OCEAN2020, 43 partners from 15 EU countries are integrating flying, floating and diving unmanned systems as sensor carriers in fleet units to set up a networked maritime surveillance and reconnaissance mission of the future. Fraunhofer IOSB is contributing its expertise at various levels: From information fusion and tracking, video analysis, development of autonomous water and underwater vehicles and coordination of the Water Strider's swarm behavior to networking and interoperability to situation display and maritime situation analysis.