Artificial intelligence improves air traffic control services

To predict air traffic, optimise traffic flows and make quick decisions are stressful and difficult assignments for ATM operators working in air traffic control. At MDH a research project is being conducted in which AI (artificial intelligence) can be used to help the operators with their assignments.

The research project ARTIMATION focuses on developing explainable AI methods (XAI), to predict air traffic and to optimise traffic flows. Safety is the most important pillar for air traffic control, and no black box process can be introduced in a decision process when human life is involved. XAI deals with methods and techniques in AI where the results of a solution can be understood by humans and thereby be of use to human operators with different work assignments. This is in contrast to the term “black box" in machine learning, where not even its creators can explain why an AI arrived at a specific decision.

– In this case the information is then analysed and presented by means of transparent AI models and explanation decisions from AI systems, says Shahina Begum, Professor and Deputy Group Leader of the research group Artificial Intelligence and Intelligent Systems at MDH.

It’s a process that will be useful for end users as well as operators who work with air traffic control – with assignments such as predicting air traffic, optimising traffic flows and decision-making. These are complex and stressful assignments, which entail a lot of processing of content and information – assignments that are done manually today. The transparent AI models with explainable decisions will be of help for the ATM (Air Traffic Management) operators, as they are called.

MDH is coordinating the project and also leading a work package called ”Lifelong Machine Learning with Human-centred AI”.

– Here the goal is to investigate the usability of AI methods from the XAI domain to establish a secure and reliable base for decisions, says Mobyen Uddin Ahmed, Associate Professor at MDH.

The project improves functionality and reliability in AI systems

The project makes use of input from human end-users when the AI system is developed. This will improve the functionality, acceptance and reliability in AI systems in general, but will also lead to the fulfilment of global goals such as the improvement of industry, innovation and infrastructure in society.

The results from the project will also be useful for AI researchers, who can benefit from the research when it comes to the transparency, generalising and explainability of AI methods. The suppliers of social systems and technology will also benefit from the results, which will hopefully lead to AI systems that will be more communicative and reliable for the human users.

The project participants, apart from MDH, are the Italian company Deep Blue, École Nationale de l'Aviation Civile (ENAC, aviation research institute in France) and the university in Rome, ”Sapienza” (UNISAP, Italy). Furthermore members of the External Advisory Board (EAB) from EUROCONTROL – a European organisation for security in air traffic (ECTRL), Belgium, and SAAB Digital Air Traffic Solutions AB, Sweden.