Schiffstechnik | Ship Technology Autonomy – virtually real? Ship design software today is not only about finding better hull shapes and making more accurate calculations. The Conference on Computer Applications and Information Technology in the Maritime Industries (COMPIT) showed a clear trend towards smarter products capable of much more. By Felix Selzer Photo: Seppo Helle At this year’s 16 th COMPIT, taking place in Cardiff from 15 to 17 May, automation and autonomy were the prevailing subjects discussed by the international experts. While some focused on smart ships, others looked at smart solutions already at the design stage. Computers can calculate process amounts of data much quicker than humans – old hat. But now developers focus on creating not only quick but also smart software that handles an increasing number of objectives and that »thinks« out of the box. One example prominent at COMPIT was the idea to take the internal architecture of vessels – the arrangement of compartments, systems and components – into account during the design process. Looking seprately at hull shape, internal arrangements and other areas may cause problems at a later design stage. David Andrews of UCL urged for a Building Block approach to address issues in the earliest descriptions of a design study. Siebe Cieraad of Defence Materiel Organisation used a design synthesis model for the effcient design of submarines. Also based on a packing approach, their solution automatically generates »varying yet balanced« designs. Mark J. Roth and his colleagues from TU Delft developed a new scoring metric to quantitatively assess qualitative properties of different arrangements that were generated with the TU Delft packing approach. This method enabled them to directly compare concept designs. The aim is a semi-automated assessment of designs. Picking the best design from tens or hundreds of possible designs »by hand« is time consuming. Also Ted Jaspers from TU Delft thought, why not let an algorithm do some of the work? He and his colleagues We use 21st century tools but 19 th century methods« Rachel Pawling, UCL came up with a clustering algorithm to bring order to ship designs generated by the TU’s packing approach. The aim was to find »families« of designs. The approach still has to be developed further, but the clustering already allows looking at designs from a different perspective. »We have hundreds or thousands of design variants with hundreds of features. When we manage to break them down into ten to 50 families, a naval architect can look at them in an effcient way,« Jaspers said. Creating a smarter design process was also the topic of Greta Levišauskaitė and her colleagues from NTNU and Ulstein Design & Solutions. They looked at ways for effcient 3D modelling while keeping control of the product lifecycle at the same time. Thus, they successfully tried a 4 th Generation Design (4GD) approach in shipbuilding, combining design and product lifecycle management (PLM) software in one environment. »Instead of a parent-child relation between components, nothing is fixed in 4GD. Each new design step can cause changes in previous designs,« she explained. Rachel Pawling of UCL wanted to »expand the scope of early stage computer aided design.« Instead of using bespoke software solutions, she and her colleagues propose using ubiquitous desktop computers to applied methods from outside naval architecture to make wider use of queuing theory, real options theory online tools and virtual reality (VR). »We use 21st century tools but 19 th century methods,« Pawling said. Henrique Gaspar of NTNU called upon his ship design peers to use Java- Script for their design solutions. The programming can be done in web browsers and compatibility with older version is secured. Multidisciplinary and multi-objective – this could be used to describe a current trend in ship design. Dejan Radosavljevic of Siemens PLM Software promoted his idea of a digital twin, complexity of modern vessels increases, saying: »We witness a shift from products being mechanical devices to being smart systems.« Simulation strategies should become predictive instead of be- 66 HANSA International Maritime Journal – 154. Jahrgang – 2017 – Nr. 7
Schiffstechnik | Ship Technology ing reactive and consider the scope of interaction between subsystems. Patrick Müller of Siemens gave an interesting example of how the idea of a digital twin works in reality, when he ended his presentation on data mining and the future of condition monitoring with a demonstration of Siemens’ eSiPod monitoring service that tells the engineer what is wrong and what to do and where to find the parts in question (read more in HANSA’s September edition). Not only software but also systems are getting smarter. Carl Hunter of Coltraco Ultrasonic looked into possibilities of improving fire safety on ever larger and complex vessels by applying sensor technology. With a growing trend towards automation, sensor technology for system monitoring becomes increasingly important. But the amount of data and the security of connections is still a major concern. Anna Lito Michala and her colleague proposed a smart wireless system condition monitoring data. Pre-processing reduces the amount of actually transmitted data, and the smart system can prioritize more important data. Scott Patterson of Babcock asked himself, how to deal with increasing complexity on board vessels. »Should we »A truly autonomous vessel must always know what to do« Azriel Rahav, Totem Plus hire smarter people? We already hire the smartest people. So we need advanced systems to make them even smarter,« he said. He presented a study on secure wireless options for ships, noting that it is not necessarily about cyber threats, but about data confidentiality, integrity and availability. But still the human is the weakest element in the data security equation, Patterson stressed. Human error is often used as an argument pro high-level automation or autonomy of ships. For some applications, unmanned operation is the only option. Fraunhofer Institute currently takes part in a Shell contest for the development of a fleet autonomous surface and underwater vehicles (ASV and AUV) for ocean floor mapping. Apart from the competition, presenter Gunnar Brink noted that the project can show how such developments of new technology can be made in an effcient way. The aXatlantic project of Heinrich Grümmer – an idea born »over a beer with friends« – intends to send an unmanned vessels across the Atlantic. But one that actually will arrive on the other side. Grümmer went for simulation-driv- Smart underwater robotics win COMPIT Award 2017 Italian maritime robotics expert Marco Bibuli was honoured for his work on team-capable, smart robots with the DNV GL COMPIT Award 2017. Bibuli works at the Italian research centre CNR-ISSIA in Genoa. Through a combination of machine vision, machine learning and cybernetics (control engineering), he has developed the CADDY (Cognitive Autonomous Diving Buddy) through to the first field testing stages. The prototypes give a glance at a future where robots become ubiquitous helpers, used for underwater inspection, cleaning, rescue operations, tank surveys and other tasks, supporting divers for increased safety and effciency. »The robot guides the diver to points of interest or executes risky tasks,« Bibuli explained. »It observes the diver to monitor his physical condition. Both communicate using standardized gestures.« Therefore, Bibuli and his team have developed a grammar that enables humans and robots to communicate in complex sentences. »The jury singled out Marco Bibuli, because his work brought advanced techniques of Artificial Intelligence to practical applications, representing innovative engineering at its finest. In Bibuli’s applications, the robots are not replacing the human, but act as caddies, supporting a diver in his work, for enhanced effciency and safety,« said COMPIT organizer Volker Bertram from DNV GL. LTR: Christian Cabos (DNV GL), Marco Bibuli and Volker Bertram »Five years ago with robotics we just represented a niche in the Compit program, now robotics have become the top topic,« Bibuli comments on the fact that human-machine interaction was subject of many presentations and discussions. He is convinced that pieces of his work such as gesture recognition can soon be used in real-life applications. fs Photo: Felix Selzer Photos: CNR-ISSIA HANSA International Maritime Journal – 154. Jahrgang – 2017 – Nr. 7 67
