The high levels of safety and resilience in air traffic management need to be maintained and optimally improved. As the next generation of air traffic management systems is pushed more and more towards digitalisation, combing these two goals remains a priority for air traffic service providers.

SafeOPS investigated how artificial intelligence (AI) solutions can enable safety applications that create a proactive, data-driven approach to safety, capable of predicting potential hazards in real-time. The project focussed specifically on developing an AI-solution decision support tool to warn air traffic controllers about the occurrence of go-arounds. A go-around prediction tool provides air traffic controllers with greater situational awareness, alerts others, and enables coordination when necessary. Thus, AI supports controllers in making more informed decisions when handling go-arounds,  avoiding knock-on effects like radar or wake separation challenges, thereby making aviation safer and more resilient.

Alongside the possible positive impacts, the provision of non-deterministic information can also introduce risks, like false predictions or overconfidence. Therefore, SafeOPS developed a risk framework to investigate the challenges of how probabilistic information could best be integrated into existing processes in air traffic management, along with associated risks, arising through the use (and possibly misuse) of uncertain information in air traffic control, with special focus on human factors.

SafeOPS produced a data pipeline that allows the processing of freely available data to investigate the landing phase of aircraft. The SafeOPS prototypic tool for go-around predictions shows promising results with further research underway to determine whether the achieved levels of accuracy are enough to impact safety and resilience in a positive manner. The project concluded by publishing guidance on AI-driven decision support tools,  based on the project’s learnings. 

Based on operational data (FDM, ADS-B, Metar, ...), SafeOPS is developing a tool chain for go-around predictions to support air traffic controllers in their decision-making. A risk model is also being provided to adequately interpret the risks associated with a go-around in the given context.

In order to pave the way towards the implementation of AI-driven solutions, SafeOPS evaluates its contribution to the safety and resilience of ATM systems. A special focus is placed on the interaction among humans (controllers) and technology within the socio-technical system.

Benefits

• Increased aviation safety and resilience
• Analysis of AI tools in a high-reliability environment
• Increased digitalisation of air traffic management