Maltese scientists teach computer how to make airplanes land safely

Data of thousands of flights ‘learned’ by computer could make flying safer and more efficient for airlines

The research team is working towards extending the use of the algorithm beyond the approach and landing phases, and onto multiple airports, to allow the computer to identify critical flight parameters itself.
The research team is working towards extending the use of the algorithm beyond the approach and landing phases, and onto multiple airports, to allow the computer to identify critical flight parameters itself.

A research team at the University of Malta has developed a computer programme that can help aeroplanes land safely by learning from the data of thousands of flights from a NASA database.

The research project Smart Flight Data Monitoring (SmartFDM) by the Institute of Aerospace Technologies uses machine learning techniques to analyse the recorded data and allow the computers to learn automatically –  without much direct human intervention or assistance – high volumes of data at a fast rate.

The scientists hope that through the processing of flight data monitoring (FDM), they can improve an airline’s operational efficiency and even predict possible safety issues before it becomes a problem.

“When flights whose patterns that are abnormal or inconsistent with the rest of the data set are highlighted, these are investigated further, giving the possibility to identify the root of the problem. The research team at the University of Malta behind the project found that this approach makes flight data analysis more efficient and more predictive in nature,” the University of Malta said.

All commercial aircraft carry flight data recording devices which capture a multitude of flight parameters at various frequencies over the duration of the flight. Typically, flight data recorded on board the aircraft are downloaded and analysed off-line by airlines, primarily as a means of monitoring the performance and operation of their aircraft. Timely analysis can help improve flight efficiency and reduce fuel consumption and carbon footprint, as well as maintenance and insurance costs.

The project is currently halfway through its set timeline, but there are already preliminary results showing a proof of concept using an existing NASA database of thousands of flights. The algorithms have been able to highlight abnormalities within the approach of flights going into a particular airport.

The research team is now working towards extending the use of the algorithm beyond the approach and landing phases, and onto multiple airports, to allow the computer to identify critical flight parameters itself.

“It is hoped that this research will also be applicable to the local context thereby making approach and landing in Malta safer, perhaps allowing the national airline to benefit from this technology,” the University of Malta said.

This project is led by the Institute of Aerospace Technologies in collaboration with the Faculty of Information and Communication Technology within the University of Malta, and QuAero – our industrial partner.

The Institute of Aerospace Technologies brings together one of the strongest and largest focused research activities of the University – that in the field of aeronautics which, in the past decade has attracted nearly €4 million in contractual research funding to the University.

Through these activities, the Institute has developed a strong and diverse portfolio that ranges from design of safety critical-surveillance systems to airborne Human-Machine Interface (HMI) design, flight operations, trajectory optimisation, air traffic management applications and support activities such as safety assessment, design of experiment, simulation, test and evaluation.

More in Technology