Designing and developing artificial intelligence (AI)-based systems that can be trusted justifiably is one of the main issues aviation must face in the coming years. European Union Aviation Safety Agency (EASA) has developed a user guide that could be potentially transformed as means of compliance for future AI-based regulation. Designers and developers must understand how the learning assurance process of any machine learning (ML) model impacts trust. ML is a narrow branch of AI that uses statistical models to perform predictions. This work deals with the learning assurance process for ML-based systems in the field of air traffic control. A conflict detection tool has been developed to identify separation infringements among aircraft pairs, and the ML algorithm used for classification and regression was extreme gradient boosting. This paper analyses the validity and adaptability of EASA W-shaped methodology for ML-based systems. The results have identified the lack of the EASA W-shaped methodology in time-dependent analysis, by showing how time can impact ML algorithms designed in the case where no time requirements are considered. Another meaningful conclusion is, for systems that depend highly on when the prediction is made, classification and regression metrics cannot be one-size-fits-all because they vary over time.

Artificial intelligence (AI) is a cutting-edge technology that can replicate knowledge, operation and, at some point, understanding at a human-like level. The AWARE project aims to develop an AI assistant application (ASA) designed to support air traffic control (ATC) operations by building a platform based on enhanced artificial situational awareness. One of the pillars of the ASA system is to develop a set of functionalities that mimic the behavior of human actions based on the development of technical tools. Regarding safety issues, conflict detection and resolution (CD&R) is the pillar to identify conflicts and avoid mid-air collisions. The goal is to build a CD&R that can be embedded into the ASA system and generate outputs that can be usable and valuable for ATC. CD&R tool is based on two subsystems: The CD component identifies potential separation minima infringements, while the CR module produces explainable resolution maneuvers with standardized syntax for seamless ATCO integration. CD uses a deterministic algorithmic approach grounded in trajectory prediction models, while CR implements a hierarchical decision-making architecture that emulates expert ATCO cognitive processes within a client-service paradigm where pilots serve as end-users.

Artificial intelligence (AI) is one of the most promising technologies that are being implemented in different areas. In particular, air traffic control (ATC) is undergoing a breakthrough with the introduction of AI for automation purposes. However, the use of AI in safety-critical systems, which is the control of air traffic, presents several uncertainties that must be solved before its implementation. Risk assessment and hazard identification are crucial for safe integration of any system, and AI-based systems are no exception. The goal of this work is to provide insights on the risk assessment developed for a specific AI-based system for ATC and how the lessons learned can be applied for further research. AISA (Artificial Intelligence Situational Awareness) Foundation for Advanced Automation is a European project that proposes building a foundation for automating some monitoring tasks and constitutes a team with the ATC operators. Risk assessment performs a safety analysis by identifying hazards, analysing them and their risks (based on probability and severity) and providing mitigation measures. The primary risks of distributed human-machine SA are identified, and many proposed mitigation measures are related to the implementation of safety requirements during the design phase of the system; 74 hazards were identified, and more than 150 mitigation measures were proposed. The main contribution of the developed risk library is that it expands the knowledge of AI hazards for their introduction into the ATC system.