Periodic Reporting for period 1 - VesselAI (ENABLING MARITIME DIGITALIZATION BY EXTREME-SCALE ANALYTICS, AI AND DIGITAL TWINS)
Período documentado: 2021-01-01 hasta 2022-06-30
The beginning of 2020 finds Artificial Intelligence, HPC and Big Data at the forefront of digital transformation. The maritime industry is embracing them wholeheartedly due to their true potential to meet the ever-increasing and complex traffic and shipping demand for safety, performance, energy efficiency, automation, and environmental impact. Meanwhile, shipping generates extremely large amount of data every minute, which potential, however, still remains untapped due to the involvement of enormous stakeholders and the sophistication of modern vessel design and operation. The key to addressing these challenges lies in research and innovation, especially in novel algorithms, tools, and platforms within the areas of AI, HPC and Big Data to unlock the new possibilities of a diverse range of current maritime applications for vessel traffic monitoring and management, ship energy system design and operation, autonomous shipping, fleet intelligence and route optimisation, and so on. While digital transformation is progressing rapidly in all aspects of society, it is now the right time to unlock the potential of extreme-scale data and advanced technologies to address the high computational, modelling and data processing demands required for accurate modelling, estimation and optimization of design and operation of ships and fleets under various dynamic conditions in a timely manner. To tackle all the above-mentioned challenges and opportunities, VesselAI has the objective to develop, validate and demonstrate a novel holistic framework based on a combination of the state-of-the-art HPC, Big Data and AI technologies, capable of performing extreme-scale and distributed analytics for fuelling the next generation digital twins in maritime applications and beyond, including vessel motion and behaviour modelling, analysis and prediction, ship energy system design and optimisation, unmanned vessels, route optimisation and fleet intelligence.
During the first reporting period of the project, a first version of the majority of VesselAI services has been developed according to the stakeholders’ requirements and use cases elaborated through the collaboration with the pilot users. Moreover, a comprehensive pilot’s operation and evaluation plan has been developed, that will guide the pilot’s operation during the 2nd reporting period of the project and provide feedback for further enhancement, polishing, scalability, and performance of the VesselAI services as well as for the proliferation of the solution with more comprehensive services and data. A SotA analysis and considerations about the progress beyond the SotA in several areas, such as Computing Architectures, Data Storage and Management, Semantic Annotation, Enrichment and Reasoning, AI Models, High-Performance Computing and Data Visualisation for Maritime Applications. An initial set of user requirements that focus on the real needs of the maritime industry have been developed. Moreoever we ivestigated the regulatory, IPR, ethical and data protection frameworks in the context of VesselAI and developed a procedure (based on the Assessment List for Trustworthy Artificial Intelligence – ALTAI) for identifying potential ethical risks and implications of AI relating to the maritime domain, and specifically the VesselAI AI services during the implementation phase of the project. Regarding VesselAI Development a first initial technical architecture has been setup and the first version of the VesselAI data and AI pipelines has been designed and developed, the dependencies amongst the various components have been investigated and fine-tuned. In collaboration with the pilots, compiled a set of state-of-the-art techniques and methods for the maritime use cases and translated them into Hw/Sw requirements for the design of the VesselAI HPC cluster in order to support the processing of their workflows and bottlenecks. In the context of the models 13 AI models/analytics have been designed and developed of initial based on the VesselAI pilots’ requirements and needs as well as the workflows to develop the models.An appropriate methodology, accompanied by the corresponding framework, for the evaluation and validation of the project pilots have been officially described, which includes both the technical (software product quality, data quality, etc.) as well as the usage evaluation of the projected VesselAI services. In terms of dissemination vesselAI made an establishment of a full online presence, including social media channels and a website allowing for the presentation of results and the on-going participation of partners, related project and broader community, VesselAI had regular event participation and presentation to through wide-purpose conferences and workshops, but also in smaller clustering and collaboration forums, where we have active synergies, including the AI4Europe / AIOD and BDVA/DAIRO community. A first draft of Business Model Canvas and partners’ individual exploitation plans has been setup
The VesselAI project aspires to revolutionise the maritime industry through the combination of HPC, big data and AI technologies to address current challenges and special requirements lying at the intersection of extreme-data processing and vessel modelling. VesselAI will research and design a data management pipeline and semantic framework to address key limitations associated with extreme-scale data processing and management, especially for distributed and streaming data, while being fully tailored and configured to handle the different types of maritime data, On the computing aspect, the project will address the topic of HPC, big data and AI convergence.The VesselAI project proposes to address those divergences at both hardware and software levels . On the hardware aspect, the project will target the design of a HPC Architecture enhanced with not only classical acceleration technologies but also AI-acceleration technologies to bridge the gap of the HPC-AI convergence for which scalability will be the key criterion. Moreoever VesselAI aims to provide a holistic building, training and serving framework for advanced ML / DL models in order to increase the validity and accuracy of the models and offer advanced life-long federated learning while minimizing bias, anomalies and current issues such as overfitting for complex models such as the vessel digital twins.
The main target of the VesselAI solution is to exploit and combine the latest technological advancements in Artificial Intelligence (DL and ML algorithms) HPC, big data and to tackle fundamental research, technical and business challenges in the definition of high-fidelity digital twins and maritime applications that analyse and predict vessel trajectory and behaviour, vessel components behaviour for optimised design, unmanned vessels behaviour, as well as weather routing and feel intelligence. Addressing these core challenges will bring a significant impact and innovation on the vast majority of current maritime workflows, applications and systems including but not limited to Collision avoidance and safe navigation, Ship Design Energy efficiency, Vessel Traffic efficiency and port management, Ship routing and scheduling, Autonomous shipping and unmanned sea vessel systems. VesselAI will significantly contribute towards the optimisation of fuel and energy consumption and, thus, the environmental impact of the maritime industry.
The main target of the VesselAI solution is to exploit and combine the latest technological advancements in Artificial Intelligence (DL and ML algorithms) HPC, big data and to tackle fundamental research, technical and business challenges in the definition of high-fidelity digital twins and maritime applications that analyse and predict vessel trajectory and behaviour, vessel components behaviour for optimised design, unmanned vessels behaviour, as well as weather routing and feel intelligence. Addressing these core challenges will bring a significant impact and innovation on the vast majority of current maritime workflows, applications and systems including but not limited to Collision avoidance and safe navigation, Ship Design Energy efficiency, Vessel Traffic efficiency and port management, Ship routing and scheduling, Autonomous shipping and unmanned sea vessel systems. VesselAI will significantly contribute towards the optimisation of fuel and energy consumption and, thus, the environmental impact of the maritime industry.