Just how useful is artificial intelligence in air traffic management? That was at the heart of discussions on day two of the SESAR Innovation Days (SIDS) in Seville on 28 November. Experts headlined some of the operational use cases where artificial intelligence applications are being trialled through research and innovation, as well as some of the challenges facing acceptance and implementation of this promising technology.
Hosted at the University of Seville, the SIDs plenary was moderated by Ruben Flohr, ATM Expert, SESAR 3 JU and brought together a lineup of experts in the field of artificial intelligence and automation in ATM, namely, Giuseppe Contissa, Professor, European University Institute, Jose Manuel Cordero, CRIDA/ENAIRE, Luis Barbero, GATCO Director, Heathrow Approach Air Traffic Controller, Paula López-Catalá, Programme Manager, Innaxis Research Institute, and Ramon Dalmau-Codina, Data Scientist, EUROCONTROL (Winner of SESAR Young Scientist Award 2017)
ATM is an ideal candidate for greater automation and augmentation through AI. With their repetitive procedures generating huge amounts of data, aviation and ATM can make use of AI and higher levels of automation to improve the efficiency of their operations in many ways and allow human operators to focus on safety-critical tasks.
Here are just some of the use cases where AI is being trialled through SESAR research and innovation:
Airport / Tower surveillance
Taxiway inspection (i.e. bird hazard, presence of drones, drones and the need for drone protection, autonomous vehicle monitoring, human intrusion, etc.) and runway monitoring (approach and landing) misalignment warning.
Read about the TRUSTY project
Traffic hotspots
- AI-based flow management Position (FMP) function to predict and resolve traffic hotspots.
- Automatic support for hot spot analysis and resolution, integration of constraints and dynamic airspace configuration (DAC). Data driven trajectory prediction.
Examples of projects addressing this use case: ASTRA, HARMONIC, ARTIMATION, DART
Network state monitoring
Prediction and management of network critical states and degraded performance
See SESAR Solution Collaborative network performance management (PJ.09-W2-49)
Smart sectorisation (SMARTS)
Dynamic airspace configuration and the design of “smart sectors”. This covers the design of basic volumes of airspace with optimal distribution of workload, tailored around specific safety and operational requirements, including complexity.
Read about the SMARTS project
Optimised runway delivery
Enhanced optimised separation delivery with machine learning uses more accurate predictions of final speed profiles derived from advanced big data / machine learning techniques.
See SESAR Solution: Enhanced optimised runway delivery for arrivals (eORD) with machine learning (PJ.02-W2-14.6a)
Controller and pilot decision support
Various AI solutions solutions (digital assistants) to support Pilots, ATC operators and Airport operators in non-safety and safety critical operations (Fast-track - exact functionality not yet defined)
Examples of projects addressing this use case: TAPAS, MAHALO, AISA, JARVIS, DARWIN
Dynamic reconfiguration of airport resources
Airport-airport coordination in strategic and pre-tactical phases based on airport stakeholders and network requirements, including both information and predictions.
Read about the FASTNet project
Improved adverse weather forecasting + impact on network management
Integration of AI-based convection prediction models within air traffic flow management (ATFM) operational tools. Improve prediction of additional weather phenomena impacting aviation (turbulence, low visibility, high altitude ice crystals, SO2 and dust). More precise characterisation of demand and capacity imbalances due to convective weather.
Examples of projects addressing this use case: ISOBAR, KAIROS
Speech recognition
Automatic speech recognition (ASR) to reduce the amount of manual data inputs by air-traffic controllers (using also airspace structure and radar data)
Examples of projects and solutions addressing this use case: MALORCA, HAAWAII, Virtual/augmented reality applications for tower (PJ.05-W2-97) and automatic speech recognition (PJ.10-W2-96)
U-space
U-Space advanced (U3) ‘separation management service’
Examples of projects addressing this use case: BUBBLES, USEPE
Understanding airspace users’ preferences and behaviour
Modelling of trajectories and estimation of non-observable variables from historical air traffic data. Particular attention will be paid to the estimation of variables related to AUs’ preferences and behaviour (e.g., airline cost functions).
Read about the SIMBAD project
Passenger behaviour
Characterisation of passengers and journeys attributes, aimed to capture relationships between the target variables missing in the mobile network data (i.e., number of persons travelling together and number of bags) and the explanatory variables that are present both in surveys and mobile network data (e.g., place of residence, purpose of the trip, last mode of transport used to access the airport).
Examples of projects addressing this use case: IMHOTEP, BigData4ATM, MAIA, MultiModX
ICT security
AI based surveillance of network load distribution to detect anomalies. Usage of AI for PEN testing. Usage of AI for systems hardening. Software defined networking.
Read about the SINAPSE project
More the CORDIS Results Pack on AI in ATM
