The development and use of artificial intelligence (AI) has accelerated steadily over the past decade. This is due to three major elements:
- The collection and storage of massive data
- The increase in computing power
- The development of increasingly complex and resource-consuming algorithms and system architectures
AI is now present in many fields and is integrated into our daily activities (smartphones, conversational assistants, etc.). The aeronautical domain is also deeply involved to integrate and benefits from this new technology.
The significant and continuous growth of air traffic prior to the pandemic has led to a strong exploration of the use of AI in the aeronautical field, especially air traffic management. AI should be able to support ATM actors in the challenges related to the increasing complexity of air traffic (e.g. traffic density, new entrants) but also to meet the challenges of reducing CO2 emissions. This new technology nevertheless raises essential questions such as legal, safety and security issues. The decrease of traffic caused by COVID-19 should allow aviation stakeholders to take advantage of the coming years to overcome these real obstacles.
Use of AI in aviation
Today, many automated assistance tools are already deployed to help air traffic controllers (ATCOs). SESAR research projects are also looking at new applications and the automation of certain tasks to facilitate the daily life of ATCOs.
AI could play a key role in the following areas:
- Demand prediction: By adding AI, algorithms based on actual flight plan and trajectory data to current computational methods, it would be possible to improve the accuracy and predictability of traffic demand. This makes the process of demand and capacity balancing (DCB) more efficient.
- Optimal sectorisation: AI can provide significant support in dynamic airspace management. Indeed, the learning of AI in this domain can allow to handle a large amount of data and complexity, beyond human capabilities. In order to provide an optimal dynamic configuration of sectors by taking into account the evolution of traffic in the different areas, and improve airspace design by suggesting redefined sector boundaries better reflecting the evolution of traffic.
- Optimisation of flight plans: AI algorithms, by integrating the numerous flight plans, traffic data, weather, etc., will be able to propose optimal trajectories for aircraft and contribute in particular to reducing greenhouse gas emissions.
- 4D Trajectory: by integrating take-off information and airport activities (delays, connections), AI contribute to give a better prediction of the “date to flight” but also to support the decision-making on specific operations, such as the go-around.
- Remote Towers: AI applied to pattern recognition can be used to recognise holding points, stands, parking positions or alert the controller to particular situations.
The use of AI can be an essential enabler to increase the safety and efficiency of services provided by ANSPs, through a variety of different tools and systems. The use of AI in air traffic management can bring significant benefits, including:
- Improving safety by managing abnormal traffic,
- Improving capacity by supporting decision making
- Understanding flow changes,
- Addressing environmental issues by optimising flight plans and trajectories.
How is Coflight exploring the use of AI?
Convinced of the advantages of AI, the Coflight teams are also exploring the use of AI to improve their system. The idea is to offer an innovative system that will facilitate the daily life of controllers.
For this, three thematics are currently under study:
- Trajectory improvement: to integrate AI algorithms to enhance the Coflight trajectory prediction. Airspace users will be able to plan greener flights, with a 4D planned trajectory taking into account airline cruising level preferences in addition to the requested flight levels, and an optimised 2D route integrating as soon also weather impact.
- Cruising speed: to determine as closely as possible the airlines’ preference between cruising speeds. To do this, the algorithms will have to predict the aircraft flight parameters in order to determine whether the aircraft is flying in economy mode or faster. This data will pave the way to a better prediction of the estimated speeds in cruise phase, and thus compute a more reliable flight plan.
- Traffic predictability: from the filed flight plans, estimate the traffic per sector, in order to better feed flow management tools and optimise HR cycles. This solution will improve the predictability of traffic and thus help in anticipating possible changes to let time to provide rapid and effective operational response.
These projects are at the study and analysis stage, the long-term objective will be to apply this research to concrete cases in Coflight in order to improve it and propose an accurate and relevant tool for real-time air traffic management.