Even though a commercial passenger aircraft is still ‘flown’ by human pilots, it is the on-board automation which actually controls the aircraft for most of the flight. The level of automation is increasing with each generation of aircraft, as new systems and features are introduced to increase efficiency and safety. Consequently, airline pilots are spending less and less time flying the aircraft manually and more and more time managing various aircraft systems.
The on-board automation works perfectly well in normal conditions; however, if the aircraft enters an abnormal state, or one or more of its critical sensors or systems fail, then the automation breaks down and the pilots are left to handle the aircraft on their own. This has happened on several occasions such as Air France Flight 447 (2009) and, more recently, Lion Air Flight 610 (2018) and Ethiopian Airlines Flight 302 (2019). In the case of Flight 447, the aircraft’s autopilot disconnected (due to inconsistencies between the aircraft’s speed measurements, caused by ice in the speed sensors) and the pilots tried to recover the aircraft manually. Unfortunately, the pilots reacted incorrectly and the aircraft stalled and eventually crashed in the Atlantic Ocean.
Even in normal conditions, certain high-level (strategic) tasks are still carried out by the pilots. For instance, one of the pilots’ responsibilities is to monitor the weather situation ahead of the aircraft – by combining information from multiple sources, including the on-board weather radar and air traffic control weather reports – and to reroute the aircraft if severe weather, such as a thunderstorm, is detected. This requires the pilots to interpret the weather situation and to figure out the best way to avoid bad weather along their route.
The SMARTAP research and innovation project – a collaboration between the Institute of Aerospace Technologies at the University of Malta, and QuAero Ltd. – is developing a novel system, based on Artificial Intelligence (AI) and Machine Learning (ML) techniques, which is intended to augment existing aircraft automation systems, including the autopilot and flight management system. AI and ML are already being applied to many other aspects of our life, such as for voice and facial recognition in smartphones and surveillance devices, automatic control of driverless cars, and fraud detection in banking.
Within the SMARTAP project, the purpose of AI and ML will be to assist pilots to complete high-level tasks which are not currently automated (such as the interpretation of weather information and the calculation of weather-avoidance manoeuvers) and to recover from loss-of-control situations which go beyond the capabilities of current automation (such as aerodynamic stalls and spins). The use of AI and ML techniques will allow the proposed system to be trained and to learn to perform tasks and deal with flying situations in a similar way as human pilots. Thus, the proposed system will act as an ‘artificial pilot’ – supporting human pilots in challenging situations and, consequently, improving flight safety.
SMARTAP (R&I-2018-010-T) is a €195,000 project financed by the Malta Council for Science & Technology, for and on behalf of the Foundation for Science and Technology, through the FUSION: R&I Technology Development Programme.