LFV and IBM investigate the use of Artificial Intelligence for Swedish air traffic control
16 November 10:11
In the joint research project Advanced Autoplanner, LFV and IBM are investigating the possibility of using AI (Artificial Intelligence) for increased automation of more complex tasks in air traffic control.
A first proof-of-concept by the means of a model that provides air traffic control instructions in a Swedish Enroute sector, where it separates aircraft in an efficient and safe way, has now been developed.
The ambition is to get funding for a continuation project in the next step in order to be able to take in more of the complex conditions that exist in air traffic management, such as climbing and descending traffic and different wind conditions.
- The challenges that exist in the project are primarily linked to traceability in the Artificial Intelligence's decision-making process, we need to understand why it makes the choices it does, the human-machine interaction, safety evidence, that there are infinate traffic scenarios to potentially train on, weather, cost and redundancy, says Sofia Rydell, LFV's project manager for the research project Advanced Autoplanner.
Additional areas of development lies within the user interface for the air traffic controller. In the interface, there will be recommended instructions - which come from AI - and intended time for the implementation by the air traffic controller. The plan is now to carry out manned simulations in a research simulator where the air traffic controller provides the clearances that the model proposes.
The current project will run for another two months. The expectation is to be able to carry out the continuation project in 2021. The project has been financed by the Swedish Transport Administration.
This film presents how Advanced Autoplanner works: