Dr Kevin Willis shares University of Portsmouth’s exciting role in the AI-ARC project

T: University of Portsmouth, B: City of Portsmouth, Credit: University of Portsmouth

In this, the second of our series of AI-ARC web posts, we catch up with Dr Kevin Willis to find out about University of Portsmouth’s exciting role in the AI-ARC project and how the service UoP are developing will be used in the AI-ARC Virtual Control Room.

Can you give some background to University of Portsmouth?

The University of Portsmouth (UoP) is a top modern higher education and research institution based on the south coast of England. UoP receives good feedback from student satisfaction surveys and ranks highly in the UK research excellence framework. While maintaining close ties to the past in this historic naval city, the university is progressive and forward thinking, with one of the universities key research themes being sustainability and the environment. The university has recently been awarded a grant worth over £3 million which will bring it a step closer to its ambition of becoming climate positive by 2030.


We are measuring potential end-user opinions on the acceptance of the new technology being developed and how these views change through the project.


Can you describe in more detail University of Portsmouth’s role in the AI-ARC project?

The UoP AI-ARC team is involved with the project through leading tasks across three work packages. We are measuring potential end-user opinions on the acceptance of the new technology being developed and how these views change through the project, evaluating the individual component services of the AI-ARC Virtual Control Room in terms of what they can add to existing systems, creating a future deployment roadmap for these services at the end project, and also responsible for designing one of the specific services in the system – the risk index computation service.

What are your Individual research/ technical interests as a team?

Members of the UoP team are split across two of the university’s faculties, the School of Mathematics, in the Faculty of Technology and the School of Operations, Systems and People, within the Faculty of Business and Law. All the team are members of the Centre for Operational Research & Logistics (CORL) offering expertise in data, algorithms, system design, modelling, programming, and decision-making. Recent research produced from the UoP team has been in diverse application areas like off-shore windfarm maintenance and drone inspection technology, Arctic and North Atlantic security and safety, allocating novel robotic devices to treatment centres, and modelling support in humanitarian crisis.


The part of the project that has been both the most interesting, is also the greatest challenge, and that is the complex nature of all the interconnected elements.


What do you find most interesting or challenging about University of Portsmouth’s role in the AI-ARC project?

The part of the project that has been both the most interesting, is also the greatest challenge, and that is the complex nature of all the interconnected elements. The designing of a service to aid decision-making in a perilous maritime context based on a set of use case scenarios to demonstrate effectiveness, the communication within the main AI-ARC system, the data requirements, the multiple services all based on advanced Artificial Intelligence ideas – and all this done with the further challenge of cross-border cooperation and collaborating with the wide range of project partners. This complexity can seem daunting but offers a real opportunity to be involved in important and interesting research.

Dr Kevin Willis, Research Fellow at the University of Portsmouth

What is your involvement in the preoperational testing?

The upcoming pre-operational testing is another big step in the project and for the services being developed. After previous demonstrations to test service integration and communication the aim is now to achieve meaningful service output and gather feedback from users participating in the tests about the system performance and the individual services. UoP are involved in a key project task of reporting on the maturity of the services and their value added to current system capabilities. We will use the feedback and system output from this testing stage as part of a future technology roadmap of what is required to have these services ready for full operational use. Of course, as one of the service developers we are also looking forward to using these tests to gather more information about the performance of our risk index.


The index will offer a multi-level output measure, which will not only indicate when risk is high, but should clearly indicate as the risk levels change with time.


Can you tell us more about the service you are developing on risk computational indexes? what are the aims of the service to contribute to the overall project?

The service we are responsible for is the risk index computation service, with risk and security being a second of UoP’s key research themes. This service is not intended to identify the specific risks, those have been outlined earlier in the project, but to assess the risk level for certain identified challenges like vessel grounding, collisions, route planning, and oil spill prevention. The service will use the available data sources, together with human-factor expert information, to act as a dynamic tool as an aid to measuring situational awareness. The index will offer a multi-level output measure, which will not only indicate when risk is high, but should clearly indicate as the risk levels change with time. These risk levels will then be visualized within the AI-ARC Virtual Control Room.


We view the risk service as a useful tool as an aid to decision making alongside all other information available to the system user.


How do you think this service can complement other services?

We view the risk service as a useful tool as an aid to decision making alongside all other information available to the system user. For example, if the risk index level changes for a specific risk type this could be registered by the user who can then look at an accompanying service to see if action needs to be taken. It is hoped that in these situations the risk index can be considered as a potential early warning system, with incident likelihood together with potential severity being key aspects of the risk evaluations. This service should help decision makers to assess risk of the various scenarios, so that actions related to prevention or mitigation can be activated.

What’s next in terms of follow on projects?

With UoP being involved in the ARCSAR project identifying safety and security needs and monitoring potential innovation solutions in the Arctic and North Atlantic region, as well as the AI-ARC project where innovation is happening it is hoped we can continue in the area of safety and security research in future projects, including the Arctic, Baltic and North Atlantic maritime regions. Like many AI-ARC partners, we are very concerned, and hence interested in researching the topic of critical underwater infrastructure. As a university, we have entered into a twinning agreement with a Ukrainian University, and our team is hence in the initial stages of joint safety and security research in a couple of fields, but for now we remain concentrated on preparing for a successful pre-operational testing in the AI-ARC project.

AI-ARC Live Demonstrations dates 2023

We are excited to share the dates and locations for the upcoming AI-ARC live demonstrations in the Autumn.

Baltic Demonstration 20-21.9.2023, Karlskrona, Sweden
The focus will be on anomaly detection such as illegal fishing, smuggling or threats to critical underwater infrastructure.

Arctic Demonstration 16-20.10.2023, Reykjavik, Iceland
In the Arctic Demonstration the focus will be on major Search and Rescue incidents as well as safe navigation including ice and weather information.

AI-ARC Event hosted by Fraunhofer IOSB in December

Ahead of the upcoming AI-ARC consortium Meeting and Lab tests in Karlsruhe from the 13th to 15th of December, we caught up with Mathias Anneken and Adrian Hoppe from Fraunhofer IOSB.

Fraunhofer IOSB, © indigo

Can you give some background to Fraunhofer IOSB?

The Fraunhofer-Gesellschaft based in Germany is the world’s leading applied research organization. Prioritizing key future-relevant technologies and commercializing its findings in business and industry, plays a major role in the innovation process. A trailblazer and trendsetter in innovative developments and research excellence, it is helping shape our society and our future. Founded in 1949, the Fraunhofer-Gesellschaft currently operates 76 institutes and research units throughout Germany. Over 30,000 employees, predominantly scientists, and engineers work with an annual research budget of €2.9 billion. Fraunhofer generates €2.5 billion of this from contract research.

Fraunhofer IOSB as one of the 76 institutes, has 7 research sites in Germany and a liaison office in Beijing. With an operating budget of €70 million and about 810 employees (including about 250 students), we focus on the three core competencies that are reflected in our name. Optronics, system technologies, and image exploitation.

Optronics is concerned with the generation of light, its beam shaping, propagation, and transformation into electronic signals – i.e. in particular with technical processes to generate images (in the broadest sense) of the world around us. System technologies ensure a holistic approach: We not only develop components and algorithms but also complex hardware and software architectures. The resulting information technology systems support people on the basis of sensor data, automate certain tasks, and/or open up new paths in human-machine interaction. We set the highest standards in terms of interoperability, IT security, and data protection/privacy. Image exploitation investigates methods and algorithms to obtain information and ultimately relevant insights from the images.

As WP3 leaders can you describe in more detail Fraunhofer’s role in the AI-ARC project?

As the leader of WP3 “Technical Development”, Fraunhofer is responsible for monitoring the general implementation of the AI-ARC solution, which includes the Virtual Control Room (VCR) as well as the different services provided by the partners. Additionally, we facilitate the contact between WP2 and WP4 stakeholders to support the development of the use cases and the demonstrations. Furthermore, different components of the AI-ARC solution will be developed and integrated by the involved Fraunhofer Institutes: Fraunhofer IOSB will provide the Digital Map Table (DigLT) software as the foundation of the VCR and a probabilistic expert-knowledge-based anomaly detection system, while Fraunhofer EMI will provide a service to estimate the reliability of AI-based services.

What are your Individual research/ technical interests?

Mathias:

Coming from a background in control engineering at the Karlsruhe Institute of Technology (KIT), I always had a passion for automation. During my time working at Fraunhofer IOSB and as a PhD student at the KIT, I focused my research on situation recognition and anomaly detection mainly in the maritime domain. I was able to gain experience in multinational projects by successfully supporting projects like MARISA and OCEAN2020. Since July 2022, I’m leading the research group “Applied Explainable Artificial Intelligence” at Fraunhofer IOSB. In my group, we develop transparent AI systems which will put an end-user into the position to not only get some results out of a black-box system but rather have an explanation for the system’s results, thus enabling the usage of AI system’s critical domains. In AI-ARC, I’m more than happy to work with a motivated international team on solving our challenges ahead!

Adrian:

Already in my studies at the Karlsruhe Institute of Technology (KIT), I developed a strong interest in the topics of human-computer interaction and computer graphics. I combined the two topics in my master thesis at the Fraunhofer IOSB, by evaluating how common PC interactions could be implemented in virtual reality (VR). After that, I stayed with Fraunhofer IOSB to dive deeper into the topics of visualization, interaction, and collaboration in VR and started a dissertation on that topic. Besides research and other topics in our department of Interactive Analysis and Diagnosis, we develop the DigLT, a web- and VR-based situational display, where questions or requirements from end-users would drive forward my research interests. Thereby gained insights and new and interesting functionality that would then flow back into the DigLT software and complete the circle before starting a new iteration. Even after completing my dissertation, the possibilities of VR still amaze me and I am looking forward to uncovering more hidden potential to increase the usability and performance of current and future software systems.

What do you find most interesting or challenging about Fraunhofer’s role in the AI-ARC project?

For Fraunhofer, the biggest challenge is leading the developments which are performed by a multitude of partners to work on the same goal of making the VCR a reality. Thus getting all partners on the same page about the challenges in the maritime domain as well as the underlying used technologies is crucial for AI-ARC success. For our developments, for sure the implementation of our technologies into the VCR and the tailoring of these technologies to the end user’s needs is the most interesting part.

© Fraunhofer IOSB

Can you tell me about the upcoming lab tests? What will they consist of, and what are the aims?

The lab-based demonstration, as the first testing phase of the AI-ARC project, will take place in M16 (13-15th December 2022) at the Fraunhofer IOSB lab in Germany. An AI-ARC consortium meeting will also occur during this time. One month before the lab test, there will be an online pre-test to assure a smooth demonstration. The lab demonstration will serve to test the technical readiness of the VCR platform and will coincide with the completion of MS 11 (Data and system integration 1 – System is functional without AI).  In addition, to prepare for the integration of AI services in the near future, all partners will demonstrate that they are capable of exchanging data with each other. This requires defining the message topics and formats to be used, establishing a connection to our data exchange node, and most importantly sending and receiving data. As some systems are not finalized yet, the integration test does not target testing the systems themselves but rather checking the validity and robustness of the communication interfaces.

Fraunhofer will be hosting the next consortium meeting – What are you looking forward to about this? 

We are looking forward to welcoming the whole consortium in Karlsruhe at the Fraunhofer IOSB and thus being able to see each other in person again. During this meeting, there will be a consortium meeting as well as integration sessions and possibilities to exchange information about the developments of the individual partners.  Hopefully, the weather and the general situation will allow us to continue our discussions in the cosy atmosphere of a visit to the Christmas market in the city centre.

Christmas Market – Photo by cmophoto.net on Unsplash