Periodic Reporting for period 1 - TADA (Terminal Airspace Digital Assistant)
Periodo di rendicontazione: 2024-09-01 al 2025-08-31
TADA (Terminal Airspace Digital Assistant) aims to enhance Terminal Airspace (TMA) performance using Machine Learning (ML) and historical data from air traffic controllers (ATCOs). TADA will provide decision support and action-recommendation to ATCOs in a human-centric way.
TMAs, particularly around major or multi-airport hubs experience heavy and complex traffic. These environments could benefit from greater automation to improve capacity, efficiency, and safety.
Currently, Air Traffic Control (ATC) in TMAs relies on ATCOs identifying flights and using tools such as Arrival Manager (AMAN) trajectory predictors, and safety nets, integrated into the Air Traffic Management (ATM) system. ATCOs analyse this data, take decisions, issue instructions and update the ATM system accordingly.
However, the ATCO generated data is rarely used beyond real-time operations or post-event reviews. TADA seeks to leverage this data with ML algorithms to detect patterns, anticipate ATC instructions and deliver intelligent decision support. A digital assistant and human–machine interface (HMI) will be developed, and AMAN will also be enhanced through ML-based insights.
TADA´s objectives are to:
● Develop an AI digital assistant tool
● Develop novel HMI
● Validate TADA and the associated HMI concepts
● Gain further understanding of the right Human-AI teaming in ATC
TMAs, particularly around major or multi-airport hubs experience heavy and complex traffic. These environments could benefit from greater automation to improve capacity, efficiency, and safety.
Currently, Air Traffic Control (ATC) in TMAs relies on ATCOs identifying flights and using tools such as Arrival Manager (AMAN) trajectory predictors, and safety nets, integrated into the Air Traffic Management (ATM) system. ATCOs analyse this data, take decisions, issue instructions and update the ATM system accordingly.
However, the ATCO generated data is rarely used beyond real-time operations or post-event reviews. TADA seeks to leverage this data with ML algorithms to detect patterns, anticipate ATC instructions and deliver intelligent decision support. A digital assistant and human–machine interface (HMI) will be developed, and AMAN will also be enhanced through ML-based insights.
TADA´s objectives are to:
● Develop an AI digital assistant tool
● Develop novel HMI
● Validate TADA and the associated HMI concepts
● Gain further understanding of the right Human-AI teaming in ATC
Objective 1: Develop an AI digital assistant tool
Description: Develop an AI digital assistant tool based on historical air traffic controller (ATCO) data that supports ATCOs in their decisions to achieve the sequences proposed by an Arrival Manager (AMAN) tool as well as to develop AI-based AMAN sequences
This objective is partly achieved and is on schedule. It will be fully achieved by month 20 (April 2026), before the beginning of the final validation exercise. The following are the key activities that have been carried out in relation to this objective
State-of-the-art literature review
Data gathering and pre-processing
Development of operational concept, use cases and functional/non-functional requirements
Initial development of Reinforcement Learning (RL) agent
The deliverables/milestones that demonstrate that these tasks have been completed are
A literature review document
Data Management Plan (Initial)
Operational Services and Environment Description (Initial)
Functional Requirements Document (Initial)
Operational concept and requirement validation
The following key activities still need to be completed to fully achieve this objective
Completion of development, training and testing of the RL agent
Integration of RL agent model with the simulation platform
Objective 2: Develop novel Human Machine Interface (HMI)
Description: Develop novel HMI based on the EASA AI framework to allow human-AI teaming between the ATCOs and the solution on how the decision-support/action selection is presented (as timely, pertinent and relevant) and with embedded eXplainable AI.
This objective is partly achieved and is on schedule. It will be fully achieved by month 20 (April 2026), before the beginning of the final validation exercise. The following are the key activities that have been carried out in relation to this objective
Familiarisation with existing HMI, tools and procedures
Definition of HMI requirements
Development of HMI wireframes and mock-ups
The deliverables/milestones that demonstrate that these tasks have been completed are
Functional Requirements Document (Initial)
HMI requirements, wireframes and mock-ups are also captured in internal documents
The following key activities still need to be completed to fully achieve this objective
Refinement of HMI requirements
HMI design validation through low fidelity simulations
Software implementation of HMI
Integration of HMI with backend software
Objective 3: Validate TADA and the associated HMI concepts
Description: Validate TADA and the associated HMI concepts by conducting early testing through Human-in-the loop validation with operational staff.
This objective is partly achieved and is on schedule. It will be fully achieved by month 21 (May 2026), once all the validation exercises have been completed. The following are the key activities that have been carried out in relation to this objective
Execution of Validation exercise 1
Workshops with end users
Initial preparation of the forthcoming validation exercises
The deliverables/milestones that demonstrate that these tasks have been completed are
Operational concept and requirement validation
Experimental Research Plan (Initial)
Experimental Research report (Initial)
The following key activities still need to be completed to fully achieve this objective
Validation exercise 2
Validation exercise 3
Objective 4: Gain further understanding of the right Human-AI teaming in ATC
Description: Gain further understanding on acceptable Human-AI teaming level in the frame of TADA project and the primary considered Approach ATCO and the Sequence Manager roles
During the workshops ATCO considered the highest human-AI teaming level they would prefer for the present roles would correspond to level 1 (decision support) of the automation taxonomy in the ATM Masterplan 2025.
This objective is partly achieved and is on schedule. It will continue to be explored until of the end of the project. It will be fully achieved by month 21 (May 2026), once all of the validation exercises have been completed and the Final Exploratory Research Report is approved. The following are the key activities that have been carried out in relation to this objective
First consortium meeting with ATCOs to define the TADA concept (October 2024)
Workshop to define the Human-AI teaming and HMI requirements (January 2025)
First project advisory board meeting (February 2025)
HMI requirement and mock-up review meeting with ATCOs (June 2025).
Description: Develop an AI digital assistant tool based on historical air traffic controller (ATCO) data that supports ATCOs in their decisions to achieve the sequences proposed by an Arrival Manager (AMAN) tool as well as to develop AI-based AMAN sequences
This objective is partly achieved and is on schedule. It will be fully achieved by month 20 (April 2026), before the beginning of the final validation exercise. The following are the key activities that have been carried out in relation to this objective
State-of-the-art literature review
Data gathering and pre-processing
Development of operational concept, use cases and functional/non-functional requirements
Initial development of Reinforcement Learning (RL) agent
The deliverables/milestones that demonstrate that these tasks have been completed are
A literature review document
Data Management Plan (Initial)
Operational Services and Environment Description (Initial)
Functional Requirements Document (Initial)
Operational concept and requirement validation
The following key activities still need to be completed to fully achieve this objective
Completion of development, training and testing of the RL agent
Integration of RL agent model with the simulation platform
Objective 2: Develop novel Human Machine Interface (HMI)
Description: Develop novel HMI based on the EASA AI framework to allow human-AI teaming between the ATCOs and the solution on how the decision-support/action selection is presented (as timely, pertinent and relevant) and with embedded eXplainable AI.
This objective is partly achieved and is on schedule. It will be fully achieved by month 20 (April 2026), before the beginning of the final validation exercise. The following are the key activities that have been carried out in relation to this objective
Familiarisation with existing HMI, tools and procedures
Definition of HMI requirements
Development of HMI wireframes and mock-ups
The deliverables/milestones that demonstrate that these tasks have been completed are
Functional Requirements Document (Initial)
HMI requirements, wireframes and mock-ups are also captured in internal documents
The following key activities still need to be completed to fully achieve this objective
Refinement of HMI requirements
HMI design validation through low fidelity simulations
Software implementation of HMI
Integration of HMI with backend software
Objective 3: Validate TADA and the associated HMI concepts
Description: Validate TADA and the associated HMI concepts by conducting early testing through Human-in-the loop validation with operational staff.
This objective is partly achieved and is on schedule. It will be fully achieved by month 21 (May 2026), once all the validation exercises have been completed. The following are the key activities that have been carried out in relation to this objective
Execution of Validation exercise 1
Workshops with end users
Initial preparation of the forthcoming validation exercises
The deliverables/milestones that demonstrate that these tasks have been completed are
Operational concept and requirement validation
Experimental Research Plan (Initial)
Experimental Research report (Initial)
The following key activities still need to be completed to fully achieve this objective
Validation exercise 2
Validation exercise 3
Objective 4: Gain further understanding of the right Human-AI teaming in ATC
Description: Gain further understanding on acceptable Human-AI teaming level in the frame of TADA project and the primary considered Approach ATCO and the Sequence Manager roles
During the workshops ATCO considered the highest human-AI teaming level they would prefer for the present roles would correspond to level 1 (decision support) of the automation taxonomy in the ATM Masterplan 2025.
This objective is partly achieved and is on schedule. It will continue to be explored until of the end of the project. It will be fully achieved by month 21 (May 2026), once all of the validation exercises have been completed and the Final Exploratory Research Report is approved. The following are the key activities that have been carried out in relation to this objective
First consortium meeting with ATCOs to define the TADA concept (October 2024)
Workshop to define the Human-AI teaming and HMI requirements (January 2025)
First project advisory board meeting (February 2025)
HMI requirement and mock-up review meeting with ATCOs (June 2025).
This report covers the first 12 months of a 30 month project. As such only partial results exist so far as only 1 of the 3 validation exercises have taken place so far
Partial results
Functional requirements
TADA comprises 3 technical sub-systems
Trajectory Calculation System – Generates 4D proposed trajectories (4DPTs) for arriving flights, using AI and machine learning to optimize flight paths
Trajectory Monitoring System – Tracks actual aircraft positions versus proposed trajectories and detects significant deviations
TADA Human-Machine Interface (HMI) – Displays 4D trajectories, alerts, and recommended actions to ATCOs offering explainable AI support
The following is a summary of the key requirements per sub-system
Trajectory Calculation
- Ingest AMAN and surveillance data
- Compute 4DPTs aligned with AMAN’s sequencing objectives
- Ensure trajectories are safe, eco-efficient, and within defined constraints
- Provide waypoints, target times, speeds, altitudes, and distance-to-go
Trajectory Monitoring
- Compare actual vs. intended positions
- Detect and notify significant deviations
- Support ATCOs in corrective decision-making
TADA HMI
- Display trajectories, distane to go, and deviation alerts
- Indicate “time-to-act” for issuing clearances
- Explain rationale behind trajectory choices
Further results will be produced in the next reporting period
Partial results
Functional requirements
TADA comprises 3 technical sub-systems
Trajectory Calculation System – Generates 4D proposed trajectories (4DPTs) for arriving flights, using AI and machine learning to optimize flight paths
Trajectory Monitoring System – Tracks actual aircraft positions versus proposed trajectories and detects significant deviations
TADA Human-Machine Interface (HMI) – Displays 4D trajectories, alerts, and recommended actions to ATCOs offering explainable AI support
The following is a summary of the key requirements per sub-system
Trajectory Calculation
- Ingest AMAN and surveillance data
- Compute 4DPTs aligned with AMAN’s sequencing objectives
- Ensure trajectories are safe, eco-efficient, and within defined constraints
- Provide waypoints, target times, speeds, altitudes, and distance-to-go
Trajectory Monitoring
- Compare actual vs. intended positions
- Detect and notify significant deviations
- Support ATCOs in corrective decision-making
TADA HMI
- Display trajectories, distane to go, and deviation alerts
- Indicate “time-to-act” for issuing clearances
- Explain rationale behind trajectory choices
Further results will be produced in the next reporting period