Towards an Automated and exPlainable ATM System
In a nutshell, the scope of this research was the systematic exploration of AI/ML solutions towards increasing levels of automation in specific ATM scenarios, through analysis and experimental activities, with the objective to deliver principles of transparency, enabling the application of AI/ML supported automation in ATM. Specifically, TAPAS has:
- Described and analysed in detail two operational cases: Conflict Detection & Resolution applied to Air Traffic Control (ATC, tactical), and Air Traffic Flow Management (ATFM, pre-tactical).
- Developed eXplainable Artificial Intelligence (XAI) methods, addressing the requirements of both operational cases, which focus on the needs of operators (and potential other actors) concerning the quality and transparency of solutions generated by XAI methods.
- Applied Visual Analytics techniques to assess and enhance explainability of AI/ML systems in ATM.
- Run experiments that assessed the applicability of XAI methods in the various levels of automation considered, exploring different ways of interaction and information exchange. The objective was to understand how operators (and potential other actors) increase their trust to XAI methods.