A.3. Automation roadmap
This roadmap covers automation functionalities that need to be developed and deployed to enable human–machine teaming, as outlined in the vision. The aim is that operations in certain phases of flight will be fully automated, whereby automation is capable of managing both nominal and non-nominal situations. In this new paradigm, the role of humans will evolve significantly, focusing on the tasks or situations too complex for automation to handle, teaming up with automation to address increasing traffic complexity. Voice will no longer be the primary means of communication, as routine tasks are managed through machine–machine applications.
For simple tasks and situations, high levels of automation are achievable without AI, while it is expected that the automation of more complex tasks will require AI. The EASA artificial intelligence roadmap 2.0 provides an AI trustworthiness framework for enabling readiness for use of AI in aviation, defining six levels of AI based on the level of human agency and oversight of the AI-based application.
The targeted automation levels for ATM are expressed as future system capabilities, which should gradually enable air traffic service providers to handle flights in automated ways within a predefined scope. Outside of this scope (when a task or a situation becomes too complex for automation to handle), automation will request the human operator to supervise its operation, as illustrated in Figure 21. For each level of automation, the figure indicates the applicable AI level of EASA, where the level of automation is achieved using AI.
Figure 21: Levels of automation taxonomy and correspondence to EASA AI levels
The human actors in the roles and functions of the ATM system will evolve in line with the paradigm shift envisaged by the implementation of the vision. New roles and functions will emerge; existing roles and functions will change, in some cases radically, whereby humans become ‘system components’ in joint and distributed cognitive systems, and humans and automation share responsibility for ensuring safety and efficiency. Automation support will also make it possible for controllers to provide services across a wider geographical area than was previously possible, with controllers’ competence being linked to their ability to operate the system efficiently rather than to specific sectors. The controller licensing schemes, and the ATM system certification framework, will evolve to support this new way of operating. Automation will be developed to the extent that situations can be handled as well as or better than when the human is involved, including full decision-making and action implementation.
The different levels of human roles are as follows.
Enhanced decision-maker (level 1). the human makes all decisions based on the appropriate overviews of all feasible options (e.g. solution space) provided by automation.
Director (level 2). The human evaluates the optimal solution provided by automation and improves it where necessary. The human has the final say, while automation performs all the calculations necessary to support decision-making.
Supervisor (level 3). The human decides which tasks/situations are to be managed by the automation and by themselves. For instance, the human air traffic controller decides which aircraft should be guided by automation. The human controller oversees and can override automation once a system’s decision is not deemed appropriate due to a particular operational understanding that is not known to automation.
Safeguarder (level 4). The system operates fully autonomously under the supervision of the human. When the system identifies that it is at risk of operating outside of its allocated operational design parameters, it suggests moving back to level 3 or lower.
The design of the next generation of ATM systems in this highly automated environment will aim to achieving a human–machine teaming. The trade-off between augmentation and assistance will be carefully balanced to avoid information overload. Technology-specific and operational metrics will be used to track system performance in both the short and the long term, with particular attention being paid to the early detection of degradation modes through leading indicators.
The ATM workforce will be involved in the design of the new ATM platforms from the early stages. System development will incorporate the well-established SESAR methodology to include iterative verification and validation, in which systems are tested by end users in isolation as well as in an integrated setting, and all relevant human performance metrics are carefully monitored. Service orientation principles will be applied at all levels to facilitate the updating of the system in both its development and its operational phases, allowing the incorporation of user feedback and providing flexibility to address emerging operational needs. Human performance expectations and responsibilities will be clearly identified and commensurate with human capabilities and limitations. The integration of AI in ATM systems will be designed to be interpretable to operators. Human competence schemes will evolve to ensure that controllers and air traffic safety electronics personnel (ATSEPs) gain and retain the appropriate skills, including relevant expert-user-level understanding of AI methodologies.
Vision and key milestones for automation
By 2030, ATM in Europe will operate at automation level 0, with progressive introduction of AI, in particular machine learning applications for supporting ATFM, ATC operations or airport landside processes (Figure 22). Significant progress will have been achieved on the development of future en route, TMA and airport platforms designed for automation level 2 and above.
By 2035, ATM in Europe will operate at automation level 2 thanks to the implementation of increased automation support tools and the transition to TBO phase 2. This will have been achieved with the implementation of sector team configurations, automatic speech recognition, user profile management, attention guidance and trajectory prediction tools supporting the early detection and resolution of potential conflicts.
By 2045, for certain phases of flight, ATM in Europe will operate at automation level 4 (or when needed at lower levels when outside of a predefined scope for level 4) thanks to the implementation of new platforms designed for human–machine teaming. Voice communication will no longer be the primary means of communication, as most routine tasks will be managed through machine–machine applications.
Figure 22: Automation roadmap