The only constant is change, but change is getting larger and faster. Although the principles of Artificial Intelligence (AI) and its core areas like Machine Learning (ML) and Deep Learning (DL) have been with us since the early 1960s, it is becoming easier than ever to harness them. Vast amounts of data can be gathered, accessed, stored and mined effortlessly and this enables researchers to develop algorithms that can support us in a wide range of areas in life. This piece will cover the present and future of AI research in aviation, specifically in air traffic management (ATM), which can truly change the way we tackle challenges in the industry.
Where the industry is heading?
As of April 2020, we are uncertain of where the aviation industry is headed. Widespread disruption caused by the COVID-19 pandemic has shaken the foundations of the world and the future is more unclear than ever. The capacity crisis has turned into a different kind of crisis, and while we hope it will vanish as quickly as it came, we cannot disregard the possibility of continued disruption. Within this framework, I strongly believe that harnessing the power of AI in air traffic management is one of the few targets industry players need to consistently strive for.
AI is hard to define, but one common definition is that it is a computer, that mimics human behaviour. Machine Learning is the next step in the evolution of smart algorithms; these are techniques that enable computers to learn without being explicitly programmed to do so. Deep Learning is a subset of ML, which makes the computation of multi-layer neural networks feasible. We expect Artificial Intelligence to exceed human intelligence in the future, but we are not quite there yet. However, what are the real advantages for air traffic controllers (ATCOs), how can air navigation service providers (ANSPs) or other industry players take advantage of these capabilities?
Focus areas in AI development
Some ANSPs have already started to research and develop AI-based solutions. Much of this progress is made through the Single European Sky ATM Research Joint Undertaking (SESAR JU), which coordinates key players across the industry to achieve research and development outcomes. While improving safety, efficiency and enhancing environmental sustainability are common goals of such efforts, the strategic aim of these projects is to fundamentally modernise ATM. Based on the developments over the last decade, the primary applications of AI include:
1. Traffic forecast
Large amounts of data are turned into information and eventually knowledge in aviation, and this makes it possible to explore current and future traffic patterns, thus improving capacity by optimising traffic flow. Algorithms that are equipped with AI can influence all four stages of flow management: strategical planning, pre-tactical planning, tactical planning and post-operational analysis.
Trajectory prediction is linked to tactical planning – when aircraft are already airborne – and this has been one of the most researched areas in the field. Models built with ML algorithms can predict movements and provide tactical estimations of future flight movements. This helps planners as well as ATCOs who are supported with forecasting tools to map out possible conflicts. SESAR JU is conducting extensive research is the field under Data-driven Aircraft Trajectory Prediction Research (DART) and has promising results.
For aircraft in the Terminal Manoeuvring Area (TMA), trajectory predication cannot be used because flight movements are highly dynamic, especially where vectoring is used. In this case, a new approach is needed, Estimated Time of Arrival (ETA) prediction, where similar flight paths are clustered in a database and artificial neural networks are programmed to forecast the exact time of landing for each aircraft. Experience shows that with the help of this method the time of arrival can be predicted within a 150 second accuracy thirty minutes prior to landing.
While pre-tactical and strategical planning have not yet been the focus for AI based research, recent research programs aim to cover these areas as well. Efficiency of airline flight planning can be highly improved with the help of AI. Complex decision on timing, flight path and frequency can be supported with such methods using historical flight data.
2. Air traffic management applications
AI can be used for much more than organising big data and forecasting traffic. Exciting research results show that various areas in the ATM domain can be supported with AI and prove that the fields of application have no boundaries.
Dynamic airspace configuration is a useful tool for flow managers to support decisions for when to open or close sectors. Optimal workload is very difficult to determine when it comes to ATCO work management. AI powered algorithms can show the minimum number of sectors that need to be kept open with respect to the optimised workload of air traffic controllers.
Voice recognition is one of the most highlighted areas of AI and ML application, and results show that the ATM industry can benefit from this as well. Pioneering voice recognition software is able to understand and execute ATCO commands with up to 95% accuracy. The special rules and logic of aviation English requires special software solutions and an ATM-specific approach, but automatic speech recognition is already a reality for some of the leading ANSPs across the globe.
Tower operations support is another important area of intelligent AI applications, where physical and remote towers can both be supported. Besides forecasting landings that will likely be go-arounds, camera feeds and streams are analysed by algorithms to recognize objects, follow aircraft with Pan/Tilt/Zoom (PTZ) cameras or identify foreign object debris on the runway. Moreover, infrared tools provide better visual to ATCOs in visibility limited conditions.
3. Unmanned Traffic Management (UTM)
Drones are the next big thing in aviation, and unmanned traffic cannot be managed without the support of AI based techniques. Besides protecting the approach areas of aerodromes, it is also a challenge to prepare for the increasing number of drone flights in urban airspace. Drones are already using these methods to avoid conflicts. Further, ANSPs also need to be qualified to handle thousands of them at the same time. The system that will be able to do this will require strategical conflict management, dynamic trajectory planning and the ability to detect and avoid conflicts all in real-time. The deployment of such a complex system will be the ultimate challenge for AI and ML developers.
The focus of the next industrial revolution is on AI and its applications. As ATM becomes more and more complex, automation is inevitable, but human factors are not to be left out of these processes. Future predictions by key aviation industry players show that AI will not replace us humans, as Nikunj Oza, a computer scientist and leader of the data sciences group within NASA Ames’ intelligent systems division said “So far, rather than AI replacing humans outright in aviation, AI and human experts have proven to be complementary—a partnership that can save human lives”.
HungaroControl’s approach on AI
The Hungarian ANSP is already harnessing the power of AI in order to leverage its benefits within simulation and operational environments. Covering most areas of AI application, efficient solutions emerge that help ease the challenges of the industry.
Virtual Pseudo Pilot is the ultimate software solution for ATC simulators that executes pseudo pilot tasks with the power of advanced voice recognition technology. Built with unique pilot logic the application can recognize ATCOs’ voice commands, acknowledges the command on human voice according to the rules of radiotelephony while executing the commands in real time.
The AI-based DeFOG tool is compatible with any video surveillance system, and enables controllers to guide aircraft even in dense fog.
The AI-based conflict resolution advisory tool will support drone operators to safely resolve critical conflicts, caused by the projected increase in airspace conflicts due to drone operations.
This article has been published in Air Traffic Management Magazine 2020 Issue 2. Click here and read on Air Traffic Management’s website.