Intelligent applications for aviation safety

E-PILOTS coordinator, Dr. Miquel Àngel Piera, has written an article at UAB divulga in which he details the aim, scope and challenges of our project:

EUROCONTROL’s forecast of an increase in the demand for flights that could double their number (Airbus/BTRE aircraft) together with a clear forecast of a lack of pilots, justified by estimates of 7 pilots per aircraft (considering rest and training periods), means that the main aircraft manufacturers such as Airbus and Boeing have on their agenda the development of new decision support tools that will allow them to reduce the number of pilots in the cockpit of commercial aircraft to a single pilot.

It should be remembered that the flight deck has undergone a great technological evolution from the Clipper 314 model with 5 crew positions (navigator, radio operators, flight engineer and 2 pilots) with a clear breakdown of operational responsibilities (flying, navigating, communicating and managing the system) to the current highly computerised cockpits with 2 crew members in the cockpit.

Achieving a safer aircraft with a single pilot in the cockpit is a challenge for major aircraft manufacturers such as Airbus and Boeing, as well as various multidisciplinary research groups, and E-PILOTS will contribute to a roadmap of smart applications that will reduce the pilot’s workload below a certain threshold.

The Universitat Autònoma de Barcelona is coordinating the European project E-PILOTS together with experts in Artificial Intelligence from the Computer Vision Centre, specialists in human factors from Cranfield University and the support of the company ASLOGIC with extensive experience in the development of new technologies in the aeronautical sector, to draw up a roadmap for new intelligent applications that will facilitate those tasks that require cognitive activity in decision-making.

One of the most innovative aspects is being able to recreate a single perception and understanding shared between the different actors who make decisions that are physically distributed on the ground (control centers) and in the air (pilots). The use of «Machine Learning» algorithms for pattern recognition opens up a range of opportunities in the aircraft cockpit, with a direct impact on flight performance indicators, and consequently, with an important margin of reduction in environmental impact. Some scenarios where more benefits can be obtained are precisely in air spaces with a high number of aircraft such as volumes near airports (known as Terminal Manoeuvre Area) where aircraft coexist and compete to descend with the best profile for a correct landing and with minimum fuel consumption. In this context, being able to identify behaviour patterns and anticipate sufficiently in advance the manoeuvres of nearby aircraft opens up an opportunity for pilots to anticipate the configuration of the aircraft and avoid manoeuvres that require high fuel consumption.

Beyond the correct configuration of the aircraft in the different flight phases, the E-PILOTS project considers the pilot and his operational context as the main actor, constantly analyzing the aircraft data and its context to provide the cockpit with elaborated information that facilitates a better perception and understanding of future events, as well as the projection of future scenarios according to the decisions he executes. This situational awareness is currently usually confirmed by the co-pilot, but the development of intelligent tools based on pattern identification opens up a set of improvement opportunities that are being validated in E-PILOTS by aeronautical and artificial intelligence experts and that will determine the roadmap for new technologies in future cockpits.

The rigour of the study lies in a series of previous international projects that have allowed the UAB team to implement socio-technological models in which it is possible to evaluate how the use of support tools affects an operator’s decision-making. More specifically, advanced simulation models are being reused to evaluate new systems and architectures for future airspace management that have been developed with experts from the Reference Centre for ATM Research, Development and Innovation (ENAIRE – UPM). These models are improved through sensorization by ASLOGIC to recognize the cognitive state of the pilot and to identify the time windows in which he will be more receptive to take advantage of the information elaborated considering its expiration to have enough time in case of performing a manoeuvre.

One of the most important contributions in this first phase of the project is to reduce the workload variability in the cockpit to values below a threshold by anticipating the pilot’s future actions so that he can carry them out in an orderly fashion over time, avoiding work peaks that could affect the safety levels required by international bodies.

At present, E-PILOTS is carrying out laboratory tests, but it is expected that in early 2020 it will carry out tests in flight simulators with students from the Barcelona Flight School / Aeroclub Barcelona Sabadell where they have excellent facilities and extensive experience for the training of future pilots.

Apart from the economic motivations in which some airlines see pilots as very expensive resources, it must be recognized that the main incidents in aviation are usually due to human failure, so the results of the research being carried out in the E-PILOTS project will have to be validated and certified through different accredited processes. At the level of social acceptance, one of the main problems to be overcome will be the social rejection of having a single pilot in the aircraft, for which an easier penetration in the market of cargo flights is expected.

Miquel Àngel Piera
Department of Telecommunications and Systems Engineering
Autonomous University of Barcelona

The original article (in Spanish) can be found here:

No Comments

Sorry, the comment form is closed at this time.