Issue link: https://maltatoday.uberflip.com/i/1344175
OPINION 25.2.2021 Alexiei Dingli Prof Alexiei Dingli is a Professor of AI at the University of Malta and has been conducting research and working in the field of AI for more than two decades, assisting different companies to implement AI solutions. He forms part of the Malta.AI task-force, set up by the Maltese government, aimed at making Malta one of the top AI countries in the world A ccording to the Internation- al Civil Aviation Organisation (ICAO) in 2018, 4.3 billion peo- ple worldwide travelled by plane. is figure shows that more than half of the world's population took one of the 40 million scheduled flights that year. Peo- ple feel secure because even though flight accidents still happen, statistically, it is the safest mode of transport in existence at the moment. Passengers are relieved knowing that competent pilots are han- dling the aircra, but very few realise or acknowledge the autopilot's important role. Some pilots claim that autopilots take care of up to 90% of the flight during regular trips. The pilot switch- es it on, sometime after takeoff, and it guides the aircraft through the pre-programmed route without direct assistance from him. When the plane is close to its desti- nation, it is usually switched off just before landing. In extreme cases, the autopilot can handle the arrival pro- cedure using an Autoland system, en- abling aeroplanes to land in weather conditions that would otherwise be dangerous or impossible. Even though passengers prefer to discount the autopilot's role, in reali- ty, achieving 40 million safe flights per year would be impossible without it. But today's planes are not the same as those of a decade ago. We have experi- enced giant technological leaps when it comes to today's aeroplanes. The automation onboard a large transport passenger aircraft is in- creasing rapidly, with pilots spending more time monitoring and managing aircraft systems rather than flying the aircraft directly. However, even though increased automation im- proved aeroplanes' safety and efficien- cy, it has also increased complexity. In some instances, it is even mak- ing it harder for human pilots to un- derstand precisely how the Artificial Intelligence system installed on the plane is working 'under the hood'. Furthermore, when the aircraft enters particular abnormal situations, the automation tends to break down, with the onboard pilots serving as the last line of hope. A typical example of an abnormal sit- uation is an aerodynamic stall. A stall occurs when an aircraft's wing stops generating sufficient lift to counteract the aircraft's weight. This circumstance can happen at any speed when the wing angle to the air- flow exceeds a specific critical value. Thus, the airflow over the top of the wing ceases to be streamlined and be- comes turbulent. This situation caus- es the loss of the pressure difference (generating the lift), which existed be- tween the upper wing surface and the lower wing surface. The aircraft stops flying and starts falling due to gravity. You can quickly try it out when you're driving a car. Stick your hand out from an open car window while it is moving with your palm facing down, thumb forward. Feel the air moving smoothly over it. Now rotate your palm, raising your thumb, and let the air pressure on your palm push it up. Your hand is flying! Keep turning until the upward force stops, and you've essentially cre- ated a stall situation. When an aircraft stalls, it essential- ly stops flying and begins to descend rapidly, with potentially catastrophic consequences unless the pilot takes corrective action. Even though a modern aircraft, is equipped with automation procedures to prevent a stall or warn pilots of an impending stall, it can still happen. Pilot overload due to an emergency, the failure of stall warning equipment or various combinations of errors can all mask the warning systems' effec- tiveness, and a fully developed stall may occur without the pilot's aware- ness. Because of this, pilots undergo reg- ular training to deal with such situa- tions. Nevertheless, we all remember Air France Flight 447 in 2009, where 229 passengers lost their lives, prov- ing that with all the safety features we have on board, these circumstances are still vulnerable to human error! To help out pilots, the Universi- ty of Malta embarked on The Smart Artificial Pilot (SmartAP) project to address this problem using Artifi- cial Intelligence (AI) techniques. For this purpose, we are training AI algo- rithms to take over and automatically perform stall recovery manoeuvers in various situations. This technolo- gy has the potential to assist pilots in hazardous conditions and ultimately improve flight safety. The SmartAP project is a collabo- ration between the Institute of Aer- ospace Technologies (IAT) and the Department of Artificial Intelligence (Faculty of ICT) at the University of Malta (UM), together with QuAero Ltd, a local aerospace consultancy company. IAT is leading the 36-month project, which is currently in its sec- ond year of execution. The results obtained so far are very encouraging, and the AI is already managing to perform a stall recovery in certain situations. However, there is still a long way to go before fully automating the entire process. But there is no doubt that the future of aviation lies in the use of AI systems. The complexity of the new aeroplanes is just too much for humans and only by automating most of their functionalities can we hope to make these metal birds fly safely. Even though your next flight won't be pilot- ed entirely by an AI, you can feel safe and relax now, knowing that you're not just in the pilot's able hands. But most probably, there's a computerised system, working silently in the back- ground to ensure that you have a safe and pleasant journey. SmartAP (R&I-2018-010-T) is fi- nanced by the Malta Council for Sci- ence & Technology, for and on behalf of the Foundation for Science and Technology, through the FUSION: R&I Technology Development Pro- gramme. Would you trust an auto-pilot with your life?