Evolution of the human operator role and automation – The target vision presented in the ATM Master Plan and in the EASA artificial intelligence (AI) Roadmap entails a technological evolution that will transform the way air traffic services are provided: human operators will delegate a substantial number of tasks to the automation, and both together will form a human – machine teaming able to handle an increasing traffic demand more safely and efficiently.

The research requires a multidisciplinary approach, involving safety, human performance, legal, insurance, regulatory, etc. expertise and shall be use-case driven. The objective of this research is not the development of an ATM solution with a high level of automation but, building on one or more ATM solutions (use-cases) proposing automation level 3 or 4 (human supervision or human safeguarding) based on conventional deterministic algorithms (i.e., not based on artificial intelligence) assess the evolution of the human operator role and automation.

Research shall develop a thorough state of the art of the HF impact on automation and mitigation methods that are applicable in ATM and propose standardized measurement methods to quantify the adverse impacts.

Research aims at identifying and analysing:

• How the technological evolution (degree of automation and supervisory vs. executive role) impacts the nature and frequency of human operators’ interventions/tasks, the required competencies, their states e.g., fatigue and subsequently their overall performance.
• Potential safety hazards related to the transition to an evolved human operator role, which might impact human operator cognitive skills and capabilities, and with the new role (e.g., specific to a supervisory role).
• The potential loss of sense of control by the human operator due to future technological developments, acknowledging the range of working environments and operational circumstances within Europe. The potential loss of sense of control might be related to a potential shift and further reduction of human operator tasks, resulting from future technological developments. Such a shift of human operator tasks might be expected but it is yet unclear into which direction supporting technologies develop on the medium and long-term and whether and how this might cause loss of sense of control.
• Joint cognitive systems and adaptative automation are promising developments, for which additional scientific studies are recommended because the maturity level and evidence for their effect on human operator workload and fatigue is still scarce. As these technological developments continue to evolve in the coming years, continued collaboration between researchers, technology developers, and regulatory bodies is recommended.

ATSPs could find solutions to reduce the risk of drowsiness in two opposing directions:

• by reducing automation such that bore-outs due to low task load are avoided,
• by increasing automation such that, during certain periods, human operators could relax or could execute other tasks than monitoring to avoid fatigue later in their working session. Both directions have benefits and risks and are not yet fully addressed in research.

Further research is therefore recommended to study these opposing approaches and address topics such as:

• Technological feasibility of (adaptive) automation that can intervene in ATC operations.
• (Operational) tasks to maintain human operator vigilance during periods low traffic demand.
• Means of optimal human operator engagement.

Research may consider meta-analyses and/or assessment of mitigation methods, and/or standardizing procedures, etc.

The research shall consider and complement the initial considerations of the “EASA ATCO Fatigue study” on the impact of new technologies on human operator workload and fatigue as well as the EASA’s approach on AI, as presented in the AI Roadmap. On-going work performed by project IFAV3 on increased flexibility of human operator validations is also relevant.

The results of the research shall aim at providing factual scientific data that could substantiate intervention strategies (e.g., further rulemaking, implementation support, oversight, etc.) in the field of human operator training, competence, and fatigue management, as well as in relation with the introduction of new ATM/ANS functionalities.

The output of the research will support impact assessment and future decision making by EASA on the regulatory needs associated to the deployment of the solution. The assessment shall include the consideration of legal accountability in case of an incident.

Research on human operator fatigue and rostering practices – The following research topics are proposed with the aim to further increase the knowledge and scientific evidence on human operator fatigue prevalence, causes and effects, and effective prevention and mitigation, and thereby support future decision-making by EASA. The research shall consider the “Study on the Analysis, Prevention and Management of Air Traffic Controller Fatigue” published by EASA in May 2024:

• Extend the scientific knowledge about the prevalence, causes and impact of human operator fatigue including a varied and representative sample of EU ATSPs and human operators (e.g., human operators of the oldest age group) in human-in-the-loop experiments (e.g., using simulator(s) or a highly controlled operational environment). These experiments shall:
  • Further research to identify and propose recommended bracket values of the eight roster elements maintaining the risk of human operator critical fatigue at low to moderate level; the bracket values should take into account and be correlated, if possible with traffic volumes and complexity, seasonal activities, and nominal and non-nominal (e.g. crisis) situations, beyond the results documented in the EASA study: collecting data during longer and more varied measurement periods (e.g. both summer and winter), targeting air traffic service providers (ATSPs) with specific schedules, work procedures, and variation in traffic volumes and complexity. If these criteria have an influence on human operator critical fatigue, an associated fatigue risk index should be provided.
  • Further research into the correlation and cross effects of the 8 mandatory parameters (e.g. number of maximum consecutive days vis-à- vis maximum hours per duty) as well as on the time needed to reduce/dissipate critical fatigue risks.
  • Further research on the various national labour laws in the EU and their impact on the rostering practices.
  • As far as possible, based on the above-mentioned research, identification of a methodology to calculate human operator staffing levels in ATSPs.
  • Investigate the impact on work-life balance and human operator fatigue of rostering schemes (e.g., days in advance rostering is published, flexibility for human operators to express shift preferences (e.g., to adapt to the individual circadian rhythms of morning persons / night owls), shift swapping between human operators / centralised shift swapping between individual human operators and the system, etc.). Investigate how the results of this study could be used within rostering and fatigue management systems.
  • Further collect data on the actual content of working hours in the EU ATSPs and confirm the share of operational and non-operational duties. Consider the nature of non-operational duties and measure the effect of these duties on fatigue and performance. Propose a definition of working hours and what it should or not include in view of the impact on fatigue. Finally, assess the effect of the rostering period scheme, the number of working hours per rostering period (and number of working hours per week (or month)) on (cumulative) human operator fatigue and determine the maximum number of working hours per rostering period to recommend.
  • Consider the nature of non-operational duties and measure the effect of these duties on fatigue and performance.
  • Assess the impact of new technologies on fatigue in an objective manner, while controlling for other factors (such as rostering and workload).
• Provide an updated assessment of current developments in fatigue detection technologies.
• Develop objective non-intrusive new fatigue monitoring technologies (e.g., wireless electrode electroencephalogram (EEG), speech analysis and webcam-based eye tracking, etc.) to be used in the ATC operational environment. Research shall take into consideration ethical and data privacy issues, particularly in the context of general data protection regulation (GDPR) guidelines. Future developments in fatigue detection and/or monitoring should therefore address the balance between leveraging the benefits of advanced monitoring technologies and safeguarding individual privacy by integrating robust data protection measures, ensuring compliance with regulations, and addressing ethical considerations to gain acceptance within the ATC community. As these technologies continue to evolve, ongoing collaboration between researchers, technology developers, and regulatory bodies is strongly recommended.
• Provide recommendations for the update of the SESAR human performance assessment methodology used by R&I projects in the SESAR programme to improve the consideration of fatigue at various stages of development and implementation of new technologies, including the assessment of the impact on fatigue of new concepts that make human operator role more passive/monotonous, for the manufacturers, the ATSPs and competent (oversight) authorities; in this regard assess the possible link with the Research project on the methods to evaluate the performance and impact of ATM/ANS ground equipment on human operator fatigue.

Proposals shall define mechanisms for guaranteeing the absence of conflict of interests.

The results of the research shall aim at providing factual scientific data that could substantiate intervention strategies (e.g., further rulemaking, implementation support, oversight, etc.) in the field of human operator fatigue management and working practices. Note that there is on-going work performed by project IFAV3 on increased flexibility of human operator validations.

Methods to evaluate safety requirements of ATM/ANS ground equipment and determine appropriate assurance levels – The lack of harmonised and recognised methods for ensuring the safety and interoperability of ATM/ANS system and constituents (ATM/ANS equipment) (e.g., identification of failure conditions, definition of hardware and software requirements, safety assurance of commercial of the shelf (COTS) equipment, etc.) has resulted in a significant number of different approaches applied by the equipment manufactures and air navigation service providers (ANSPs). Although there are industry standards and methods available for determining the appropriate safety assurance, these standards are not fully compatible with each other.

Furthermore, modern ATM/ANS equipment and those envisaged to by the ATM Master Plan are to make significant use of data through the application of virtual systems (e.g. through application of cloud computing).

With the transition to the EASA framework for attestation of ATM/ANS equipment (Commission delegated regulation (EU) 2023/1768 of 14 July 2023), there is a need to ensure a common approach and understanding of the safety requirements, liability aspects, assurance level and that harmonised methods are applied.

Research shall aim at providing data and information to determine:

• Certification characteristics and performance of hardware platform cloud computing and COTS solutions/equipment.
• How best to ensure the suitability for use of COTS equipment or constituents.
• Principles, assurance methods, and safety considerations to be applied in guaranteeing computing platform, virtual systems, and software applications provide their performance and safety targets.
• A methodology applicable to ATM equipment to determine “failure conditions”.
• Shared liability principles for assurance of certified equipment being used in a more highly automated operating environment.
• Principles, methods, and safety considerations to determine software assurance level (SWAL) and hardware assurance level (HWAL).

The application of airspace classification in Single European Sky airspace – Through the application of SERA.6001 Classification of airspaces of the Annex to Regulation 923/2012, a common definition of the airspace classification has been implemented. However, the designation by the Member States has resulted is an unharmonized application which leads to flight inefficiencies, decreased safety and difference in service expectations when conducting operations in similar airspace within different Member States.

Research shall provide the data and information (including U-space implementation), to determine:

• The distribution of the application of airspace classification in Member States airspace and the context of such application. The research must address in particular the implementation of class G airspace across Europe.
• A reasoned framework (including a set of parameters based on traffic demand) to support a harmonised application of the airspace classifications.

The research should consider current traffic demand and future traffic forecast, considering (in particular) VFR and IFR electric aircraft as per the EASA certification projections, as well as very low level (VLL) operations.

A harmonised application of airspace classifications in Europe will support the safe and effective operations by commercial and large aircraft and general aviation. Research shall provide the required evidence and initial inputs to define an intervention strategy (e.g., further rulemaking, implementation support, etc.) to define the classification application conditions in support of a Single European Sky.

Development of guidelines for the design of future artificial intelligence (AI) systems – Research shall aim at supporting the evolution / update of EASA guidelines for the development of AI enabled systems in ATM, including feedback on the effects of conformance, transparency and complexity and other challenges associated to the design of future AI systems (e.g., trade-offs between privacy and transparency, trustworthy AI approaches). Research shall take as starting point the issue 02 of the EASA AI concept paper.

Research shall identify concrete applications of EASA guidelines and define the appropriate activities, not only human-in-the-loop simulations considering controller trust, acceptance, workload and human/machine performance but also new approaches for validation, verification, and testing of AI applications, specifically for safety critical applications (e.g., developing an agile validation methodology and data centric security capabilities for AI systems to promote their reliability, increase trust on AI, and maintain a competitive edge in today’s rapidly evolving technological landscape).

Close coordination with EASA is expected, to ensure complementarity and consistency with EASA activities on the following areas:

• Trustworthiness: capability to keep AI-based systems with relatively high cyber-security protection. Support the definition of the requirements and needs for input/output verification (related to trustworthiness in the framework of Structured Transparency) in the ATM context in support of the EASA certification process descriptions. Validate and further develop requirements and potential solutions with a co-joint analysis together with EASA and other operational experts. Clarify some of the challenges faced by EASA (e.g., to define the system requirements, processes, and tools that are needed to perform the validation and certification process).
• Learning Assurance: including the consideration of realistic operational cases in realistic operational conditions and new machine learning (ML) techniques. Need to develop specific assurance methodologies to deal with learning processes.
• AI explainability, which goes beyond the ML techniques to extract information from the models and includes the interactions with other systems and with the human operators (human factors). Research may help to clarify which requirements and processes the target AI/ML system should comply with to be certifiable for operations.
• AI Safety case: discussing with EASA and other safety experts about the needs and requirements of a concrete safety-case can help to clarify and support the development the EASA guidelines for certification.

The concept of safety critical levels needs to be further developed for AI applications in ATM. Research covers the definition and analysis of safety-related use cases for different safety level assurances. These safety levels may imply either the adaptation of current software (SW) verification methods or the development of new ones to guarantee the safe of operation of AI in ATM.

Enhancing robustness and reliability of machine learning (ML) applications – Research aims at enhancing machine learning (ML) applications to ensure they are technically robust, accurate and reproducible, and able to deal with and inform about possible failures inaccuracies and errors. Research aims at developing potential solutions to address this challenge, which shall include/refer to the EASA methodologies for certification of AI in aviation. The research must be focused on the application of ML to ATM, by either leveraging existing ML techniques or by developing new ML techniques to address the specific challenges. Research shall consider the results and recommendations reported in the machine learning application approval (MLEAP) final report.

The scope may address:

• Further the research on “generalisation capabilities of ML models and constituents”, as the MLEAP final report indicates the need for further work (the set of methods experimented on use cases do not provide satisfactory generalisation bounds and other methods should be further investigated).
• Verification methods of robustness for machine learning (ML) applications. Due to the statistical nature of machine learning applications, they are subject to variability on their output for small variations on their input (that may even be imperceptible by a human). Research aims at proposing new methods to verify the robustness of machine learning applications, as well as to evaluate the completeness of the verification.
• Standardised methods for evaluation of the operational performance of the machine learning (ML). Research addresses the definition of reference methods and metrics to assess the accuracy or error rate of ML applications.
• Application of transfer learning and data augmentation techniques for the development of the proposed applications, thus guaranteeing their robustness. In addition, these systems would be continuously validated using ML Ops methodology and explainability techniques, to ensure system performance and detect as early as possible if concept drift is occurring.
• Identification, detection, and mitigation means of bias in ML applications. Machine learning applications are subject to bias, which can compromise the integrity of their outputs. One of the most challenging aspects when collecting, preparing, or using data, is the capability to identify, detect and finally mitigate adequately any bias that could have been introduced at any time during the data management and/or of the training processes. Research aims at developing potential solutions to address this challenge.
• ML/AI-based systems must be designed, deployed and executed while considering cyber-security aspects to prevent, detect, mitigate and respond to attacks and ensure that the system is cyber-resilient. Peculiarity in threat models, risk assessment, and monitoring of ML/AI systems must be considered.

Support to the certification of novel ATM (AI-based and non-AI-based) systems that enable higher levels of automation
The objective of this research element is to address issues related to the certification of:

• Novel AI-based ATM systems that enable higher levels of automation
• Novel non-AI based ATM systems that enable higher levels of automation

Research will address solutions, methods, etc. that could support and harmonise certification of innovative ATM systems based or not on machine learning or artificial intelligence techniques (e.g., scenario-based testing, reinforcement learning for control systems, etc.). It is expected that proposals define a holistic approach to address this challenge considering not only technical aspects of the certification but also legal and regulatory aspects including privacy. Research may explore and assess potential approaches that could be applied for the certification of automation and that allow to demonstrate the safety of automation during nominal and non-nominal conditions. Of particular interest is to show how safety can be ensured even if not all situations and variations of parameters can be anticipated during the design phase. Proposals may apply uncertainty quantification to address this issue. Research may also address the specific challenges of certification of automation that can adapt its behaviour to changes of the environment over time. Research activities shall consider other initiatives developing safety of life systems that may have different approaches to certification and review their applicability to ATM (e.g., EGNOS). Research shall consider the work performed by project HUCAN.

Development of a framework to achieve effective Human-AI Teaming – Based on the published EASA Artificial Intelligence (AI) Roadmap 2.0, the issue 02 of the EASA AI concept paper was published. This guidance document develops a novel layer of AI trustworthiness guidance related to Human Factors for AI, which is necessary to manage the approval of Level 2 AI applications, which encompasses (Human-AI Teaming). Such applications bring the level of assistance from the AI-based systems to the Human end-user one level beyond, enabling automatic decision-making or action implementation, which was not foreseen in the Level 1 AI applications (Human assistance and augmentation).

When considering an AI-based system as a part of a team, rather than simply a tool capable of limited actions, the need for a framework for improving the design of AI-based systems to enhance the overall success of Human-AI teams becomes obvious. A failure to consider the needs of the many air traffic controllers, pilots, flight dispatchers, flow managers, etc. who are responsible for successful operations will result in AI technologies that eventually fail to provide the necessary high levels of performance and may instead cause inefficiencies and safety concerns.

The design of AI-based systems for Human-AI teams needs to incorporate several highly interrelated considerations. These include designing the AI system to support not only task work, but also teamwork. These interrelated considerations include considerations about Human-AI team performance and processes, AI-based system situation representation, shared situational awareness, human team member training needs, Human-AI interaction methods, interface, AI operational explainability and Human-System Integration processes, measures, and testing.

Research aims at investigating concrete and feasible means of compliance for the new layer of Human Factors objectives and how compliance could be assessed including a definition of KPIs for performance in new roles for human, non-human, and hybrid teams. The research project could also lead to complement anticipated means of compliance for the Human-AI Teaming.

Research may include the creation of frameworks / methods for training AI-based systems together with humans, to be able to include in the objective functions notions of collaboration or KPI related to team success, and not only individual goals. The absence of standardised testbeds in AI-based ATM research fragments it and prevents truly collaboration between the research actions, even more so in the domain of Human-AI Teaming.

The research shall take as a starting point one or more use cases of application of automation level 2 to ATM that do not use AI and are already at a maturity level TRL6 or above and investigate the potential introduction of AI to enhance the performance of the Human-AI team.

Research should demonstrate a clear relationship between the human factors objectives and implementation in the wider socio-technical system (e.g., training, procedures, competence certification, etc.).

Along with the research, at least one real-scale aviation use case per domain (covering at least ATM/ANS and airworthiness) should be developed to demonstrate the effectivity and usability of the proposed methods and tools.

Explainable Artificial Intelligences (XAI) – AI explainability is the capability to provide the human with understandable, reliable, and relevant information with the appropriate level of detail and with appropriate timing on how an AI/ML application produces its results.

Applicable EASA guidance, which shall be considered by the research on this topic distinguishes between development & post-ops explainability (driven by the needs of stakeholders involved in the development cycle and the post-operational phase) and operational explainability, which refers to the need to provide end users with ‘understandable’ information on how the AI/ML-based system came to its results.

The research shall address the following aspects:

• Elaborate a state of the art review to evaluate the progress made on XAI by several research groups (e.g., DEEL (dependable, explainable and embedded learning)).
• Based on the state of the art review identify and develop further axes of research.
• Investigate the “relevance property” highlighted in machine learning application approval (MLEAP) final report[11]. The impact of inputs on outputs is an important consideration to promote when trying to explain complex models such as neural networks (NN). Similarly for control related applications (e.g., reinforcement learning), the “reachability property” from the same MLEAP report may also be of interest.
• Despite the inherent case by case nature of compliance methods to explainability objectives, it is important to research a common baseline of methods/tools for specific groups of AI/ML applications (e.g., type of technology, type of application, dimensionality, etc.).

The objective of this research is to improve transparency of automated systems in the ATM domain investigating methods based on Explainable Artificial Intelligence (XAI) in operational use cases e.g., predicting air traffic conflict resolution and delay propagation, validating the robustness and transparency of the system, etc. Research shall consider the output of project ARTIMATION and MAHALO.

Innovative methodologies for ATM safety, security, and resilience – Research aims at developing methodologies (or evolution of existing ones) for safety, security and resilience that will contribute to ensure that ATM is robust against ever-evolving risks, threats, and disruptive events in the physical and cyber worlds in a novel ecosystem (e.g., enabled by automation level 3 and above). Moreover, research shall consider how novel virtualized and distributed ATM service architecture can be cyber-resilient and collaborate to enhance the overall security approach. New and disruptive technologies, operations, and business models to ensure ATM is resilient against internal and external threats, including health, natural disasters, terrorism, and criminal activity. Research shall ensure coordination with EASA. Research shall consider the work performed under projects SEC-AIRSPACE, FARO and FCDI.

Applications of Data4Safety – Data4Safety (also known as D4S) is a data collection and analysis programme of the European Union Aviation Sector that will support the goal to ensure the highest common level of safety and environmental protection for the European aviation system.

The programme aims to provide a big data platform and analysis capability at European scale and level, including a structural link with ECCAIRS2 that enables analytics and insights from the European Central Repository safety data (ECR as per Regulation (EU) 376/2014). This means collecting and gathering all data that may support the management of safety risks at European level including safety reports (or occurrences), flight data (i.e., data generated by the aircraft via the flight data recorders), surveillance data (air traffic data), weather data, etc. As for the analysis, the programme’s goal is to help to “know where to look” and to “see it coming” as well as to support data-driven changes at system level. In other words, it will support the performance-based environment and set up a more predictive system. More specifically, the programme will allow to better know where the risks are (safety issue identification), determine the nature of these risks (risk assessment) and verify if the safety actions are delivering the needed level of safety (performance measurement).

Research aims at defining, developing, validating, and assessing potential future applications / use cases of the data collected under Data4Safety Programme, which could be later integrated during the next stages of the D4S development phase. The goal is to improve the overall capacities of the European Union aviation system to manage risks and support data-driven changes with adapted aviation intelligence, by developing the capability to discover vulnerabilities in the system across terabytes of data.

The focus should be on the utilization of training data for ATM human operators and pilots in correlation with aviation data derived from in-service operations, rotorcraft, general aviation, and drones’ operations and in the field of environment.

Automation of the security risk assessment (SecRA) process – Security risk assessment is a resource-intensive, time-consuming process which incorporates the identification of assets, vulnerabilities, threats and threat scenarios, the evaluation of risk, and the selection of security controls to meet organisational security objectives. There is currently a global shortage of cybersecurity practitioners who can do this work, and this will remain the case for the next few years.

New European regulations (Part-IS) mandate information security management system (ISMS) requirements on aviation organisations and authorities, many of which have previously not been subject to such requirements and may not have implemented an ISMS or carried out security risk assessments in the past. The main objective of Part-IS is to address information security risks which may have an impact on safety, so mechanisms must also be in place to support the coordination of the aviation safety and security disciplines.

Automating the security risk assessment (SecRA) process would assist organisations and authorities to meet the needs of Part-IS by easing the development of SecRAs while reducing the resources required.

Possible phases in achieving this:

• The automated update and maintenance of the required catalogues in an existing SecRA (e.g., assets, threats, vulnerabilities, and controls) from established sources of such data.
• The automated generation of reports on the impact of catalogue updates on an existing SecRA (e.g., describing which parts of the SecRA are potentially impacted by a new threat, a new vulnerability, a modified control, etc.).
• The development of a new SecRA, or the modification of an existing SecRA, by an information security specialist supported by an intelligent assistant.
• The autonomous development of a new SecRA, or the modification of an existing SecRA, by an AI agent.

In addition, the utilization of Intelligent Assistants (IAs) could facilitate Human/AI teaming in security and safety risk Assessment activities, such as in the following areas:

• Providing support to safety and security experts in assessing the potential impacts of security incidents on safety, and in the optimal selection of security controls.
• Assessing the potential impact of security controls on safety – and vice-versa.

Climate and environmentally driven route charging – Research shall address the potential of climate and environmentally driven route charging, with new mechanisms for charging airspace users to incentivise minimum climate impact. Route charging will reward those who avoid volumes of airspace with a high climate impact and disincentivise flight planning through high demand sectors / flight altitudes except where it optimises environmental benefit overall, while being cost neutral to airspace users and passengers on average. Added capacity in the “greener” volumes of airspace enabled by reduced vertical separations limits necessary flight plan modifications, furthering acceptance of the approach. Note that there is on-going work on this research element under projects Green-GEAR and AEROPLANE.