In the last decade marine and air traffic has increased 7-fold, energy & mineral exploitation has increased dramatically and terrorist and criminal activities have moved their activities to the sea. At the same time the maritime economic zones have been increased to 350 nautical miles from the coast. All of these factors have put a major burden on the nations responsible for managing these large busy areas. To deal with this increase of responsibility combined with downward pressure on operation budgets, cost effective situational awareness systems need to be developed that can provide solutions. These systems must be able to monitor and analyze thousands of objects and highlight any that may present a risk. Incidents at sea can often result in loss of life and cost 100s of millions of Euros.
The ICT challenge is in creating viable situational awareness systems is to develop technologies that can automatically collect, align, reason upon and visualize information from a number of heterogeneous sources pertaining to public safety and security. There is a significant amount of data available that is continuously being collected concerning moving objects (i.e. people, vehicles, aircraft or vessels). Assimilating this large and varying data set into information that can pinpoint areas of increased risk is formidable. Accuracy at all stages is essential. False information introduced early in the analysis process can give disastrous conclusions at the end.
The study case chosen by Metis is the management of large maritime areas. Here operators have the task to monitor and manage large numbers of vessels and to make informed decisions. These decisions primarily concern actions to take to ensure a safe and secure environment. Traditional and reliable sources of information such as radar are essential but if these can be enhanced with information gleaned from the internet and other sources then a whole new insight can be presented. Information and images from uncontrolled websites, social media sources and “semi-official” databases whilst being potentially valuable are not semantically aligned, complete or necessary reliable.
All partners made major steps towards realizing the goal of a system capable of understanding what was happening in a complex and dynamic environment. A 24/7 high-fidelity prototype was set up to monitor in realtime marine traffic in the Dutch Economic Zone. The system was configured to expose any vessel with an enhanced risk of smuggling, reckless behavior, risk of pollution and illegal fishing.
Biggest results so far
Visualization of topic evolution in news articles
More and more texts are available in digital form. Especially when these texts go back over many years (like with digitized newspapers and archives), it is interesting to understand how topics have evolved over time. How did a political situation change? How did a company start and evolve?
Our demo shows an automatic visualization of the evolution of a topic over time. We take a different approach on traditional search result ranking methods as for example used by Google. We show a summary over time of the topic searched for. This summary can be used to quickly get an overview of the back-story of an entity in the text. More.
ICT science question: How can we visualize the historical background of an entity appearing in a text in a concise way, without overloading the user with information? This is a hard problem, because of the need to summarize and visualize a large search result without having to cut off the information at a certain point (such as after the first page of Google search results).
Involved COMMIT/partners: Synerscope, ESI, VU.
Automatic risk assessment of vessels in maritime areas
In the last decade marine traffic has grown greatly. At the same time terrorist and criminal activities have moved to the sea as well. The nations responsible for managing maritime areas therefore more and more face the problem of detecting suspicious vessels. Our demo shows how a human operator responsible for managing large complex maritime areas can be assisted by automatic risk assessment.
Our system takes inputs from reliable monitoring systems and combines this with inormation sources of unknown trustability (such as open source intelligence, public databases, websites and social media feeds). The system automatically presents suspicious vessels to the human operator. This automatic risk assessment can give a productivity boost to the coast guard, the police, the navy, the customs and to environmental protection agencies. More.
ICT science question: how can systems collect and use information to develop a human-like understanding of complex dynamic situations? More specifically: How can systems automatically determine the intent of an action? Current so-called ‘situational awareness systems’ are characterized by cleverly visualized and configurable operational pictures supporting the decisions of the human operator. If the human is taken out of the loop, the system becomes entirely ineffective. The unique approach of our demo is the integration of multiple artificial intelligence technologies into a seamlessly automatic operating solution providing the highest end-user value.
Involved COMMIT/partners: Thales, VU Amsterdam, TU Delft, TU Eindhoven, TNO-ESI, Radboud Universiteit Nijmegen
Let an intelligent machine explain its decisions
More and more decisions that traditionally were taken by humans, are taken by intelligent machines that perform complex reasoning. An automatic pilot can fly and even land an airplane. A medical expert system can propose diagnoses and treatments of patients based on their symptoms.
In current systems, only the results of the automatic reasoning are shown to the user. This makes it hard for the user to understand and trust the results.
Our demo shows how a user of a safety and security system uses visualization to understand why and how an automatic reasoning system has reached its conclusions. This allows the user to gain both a better understanding of the situation and to improve trust in the reasoning system. More.
ICT science question: how can we best visualize the reasons why an automatic reasoning system has come up with certain conclusions? The combination of the use of probabilistic reasoning and subsequent visualization of the reasons behind the decision is new and especially important in costly and safety critical situations.
Involved COMMIT/partners: Thales, VU Amsterdam, TUDelft, TU/e, ESI, Radboud Nijmegen.