Project Bluebird: An AI system for air traffic control – sUAS News

Advancing probabilistic machine learning to deliver safer, more efficient and more predictable air traffic control

The Bluebird Project is a partnership between NATS and the Alan Turing Institute, supported by an investment from EPSRC. The vision of the research is to provide the world’s first artificial intelligence (AI) system to control part of the airspace during live tests, working with air traffic controllers to help manage the complexities of their role. This system will use digital matchmaking and machine learning technologies, and will include tools and methods that promote safe and reliable use of AI.

Air traffic control (ATC) is a remarkably complex task. In the UK alone, air traffic controllers manage up to 8,000 planes per day and give instructions to keep aircraft separate and safely. Although the aviation industry has been affected by the pandemic, European air traffic is expected to return to pre-pandemic levels within five years. In the long term, the increase in passenger numbers and the proliferation of unmanned planes will mean that UK airspace is busier than ever. Next-generation ATC systems are therefore needed to choreograph aircraft movements as efficiently as possible, ensuring the safety of our skies while reducing fuel consumption.

The project has three main research themes:

  • Developing a probabilistic digital twin of UK airspace. This physics-based real-time computer model will predict future flight paths and their probabilities – critical information for decision making. It will be trained on a NATS data set of at least 10 million flight records, and will take into account the many uncertainties of ATC, such as weather or aircraft performance.
  • Build a machine learning system that works with humans to control UK airspace. Unlike current human-centric approaches, this system will simultaneously focus on the immediate high-risk detection of potential conflicts between aircraft and on the low-risk strategic planning of the entire airspace, thereby increasing the efficiency of ATC decision making. To achieve this, researchers will develop algorithms that use the latest machine learning techniques, such as reinforcement learning, to optimize aircraft trajectories.
  • Design methods and tools that promote safe, explainable and reliable use of AI in air traffic control systems. This will involve experiments with controllers to understand how they make decisions, so those behaviors can be taught to AI systems. The project will also explore ethical issues such as the liability for errors of a human AI system, how to build a system that humans trust, and how to balance the need for safety and efficiency.

All of these themes are based on a desire to create a vibrant AI research community in this field, made possible by knowledge sharing, a peer review framework and two-way communication with TAC researchers.

The overall goal of the project is to provide the first AI system to work with air traffic controllers and control part of the airspace in live tests, which will put the UK at the forefront of technical advances in this sector. More generally, research into AI technologies in ATC will be a catalyst for scientific discovery, providing ATC with insight into new and innovative ways to modernize UK airspace, increase its efficiency and ” help the UK aviation industry. to achieve net zero carbon emissions by 2050.


The methods and theory of the project have direct applications in other scientific fields, in particular:

  • Digital twins: These transform many aspects of engineering, enabling simulation and data-driven decision making in advanced engineering systems.
  • Computational statistics and quantification of uncertainties: New advanced computational statistical methods for the calibration and assimilation of large-dimension uncertain systems will provide direct application in many fields of science that use uncertainty quantification.
  • Machine learning control and optimization: Theoretical work will explore the robustness problems in reinforcement learning and evolutionary optimization problems, as well as new results in hierarchical multi-agent systems. These outputs will provide further results in the approximation of nonlinear functions (eg, deep neural networks) and optimization methods in stochastic environments.
  • Scientific computing and high performance: The real-time nature of ATC requires extremely fast simulations and sampling techniques of a large complex system. This will require algorithmic approaches for modern high performance computing hardware.
  • IT ethics, society and AI: The project will provide new results on how social science studies can be technically integrated into new AI frameworks, and how new approaches can be developed to improve human-AI collaboration through AI methods explainable.
  • Legalization of AI control: the project will provide important data on the distribution of responsibilities within a human AI system.

Opportunities to join the team

We will recruit additional team members as the project progresses and link to the opportunities below as they arise. These two roles are the first to be recruited and will work closely with PI and partners to deliver this exciting interdisciplinary research partnership:

Project managers

Turing / University Partner Leaders

  • Professor Tim Dodwell (IP, Turing / Exeter)
  • Professor Mark Girolami (Co-I, Turing / Cambridge)
  • Professor Richard Everson (Co-I, Turing / Exeter)
  • Dr Adrian Weller (Co-I, Turing / Cambridge)
  • Dr Edmond Awad (Co-I, Exeter)
  • Dr Evelina Gabasova (Turing, Co-I)

NATS runs:

Source link

Comments are closed.