Skip to main content

FIVER feeds data to the ‘brains of an aircraft’

The worlds of big data and artificial intelligence open up new possibilities for flight management systems (FMS), and that's not too surprising, given that FMS systems consume and process vast amounts of data that contain all the coded approach, landing and departure procedures for the world's airports. 

Massive testing of Thales FMS using cloud computing and artificial intelligence
Massive testing of Thales FMS using cloud computing and artificial intelligence

The FMS also automates many different in-flight tasks oriented around performance, management of the aircraft systems, and navigation, assisting the pilot in managing the whole flight, optimising the fuel, enhancing flight safety and improving environmental mitigation. The FMS could indeed be described as the ‘brains of the aircraft.’

In Clean Sky's recently completed FIVER project the aim was to collect reference trajectories and deliver software that simulate pilot actions so as to challenge the FMS with representative flight situations and generate many computed trajectories. Factors that determine the optimal trajectory of the flight include such pilot actions but also the current status of the aircraft (position, speed, avionic configuration) relative to the phase of the mission, meteorological data, positioning of other surrounding air traffic, combined with many other operational factors.

In a nutshell, the idea is to leverage insights gleaned from massive testing to improve the awareness of all parameters and trajectory options which are relevant during flight. After optimisation by Thales, the FMS will assist the pilot in anticipating and/or identifying and executing the safest and most energy-efficient trajectory. 

By having more precise fuel management, this in turn avoids embarking on flight vectors that are suboptimal, thereby conserving fuel and reducing CO2 and NOx emissions.

'Massive testing' enables the simulations to be run in parallel at a massive scale and at very accelerated speeds thanks to cloud computing, making it possible to simulate several thousand flights a day to test the FMS operation.

FIVER brings together three partners: Thales AVS, the developer and owner of the FMS, plus two French companies. These are CGX AERO, specialised in procedure design, cartography and aeronautical software development, the co-developer of the simulation system with the support of Thales AVS. 

The other company is Scalian (formerly Eurogiciel), specialists in big data and data science techniques, responsible for bringing the data science solution to analyse the FMS massive simulations. 

The project contributes to Clean Sky's Disruptive Cockpit project within the Large Passenger Aircraft initiative, under Topic Manager Thales. 

But there's a broader benefit, says Clean Sky project officer Alexandra Gurau: ‘FIVER is a cross cutting project as it has applicability not only to Clean Sky's Large Passenger Aircraft initiative but also to our Systems initiative. It's a good example of the synergies and cooperation between the two integrated technology demonstrators.’ She adds that, given the focus of the project on autonomous aircraft trajectory management and on functions, ‘that's where this cooperation between the Large Passenger Aircraft initiative and Systems falls into place.’

‘FIVER is a good example of the synergies and cooperation between the two integrated technology demonstrators’

‘Today, the trajectories to be flown, at departure and arrival of airports, are subject to three major steps: procedure design, coding and on-board generation within the avionics (i.e. FMS). These are independent processes, so that's why, at an overarching level, we've been working on an end-to-end validation of all these three steps – this has never been previously performed,’ Gurau says. ‘The benefits, ultimately, are improving safety by securing all of this whole data chain from originator to the embedded FMS.’

The simulation environment

To realise the project objectives, the FIVER consortium put together what project manager and data scientist Mathieu Damour at Scalian describes as ‘a combination of open-sourced and protected software, allowing a near open world experience of flight simulation.’
‘The CGX AERO and Thales AVS simulation environment includes the virtual representation of real aircraft and aircraft pilots with several levels of realism,’ explains Damour. During the simulation, additional contextual flight data are enriching the FMS computed data, forming the pool of data that will be analysed.

The simulation environment includes the virtual representation of real aircraft and aircraft pilots with several levels of realism

By integrating pilot’s cockpit interactions it's possible to virtually reproduce realistic environments using real meteorological data and using real data from airports, pilots, aircraft and FMS systems.

‘In a given situation, at each moment of the virtual flight the virtual pilot will be able to replicate various pilot/cockpit interactions such as the activation of an avionic system (ex: auto-pilot, managed guidance…), or the revision of flight data and directive given to the FMS, FCU (flight control unit) and cockpit, or the modification of the flight plan with simulated ATC (air traffic control) involvement,’ says Damour. In FIVER, a set of flights as they might have occurred in reality was played out on a large scale and the behaviour of FMS, among other things, was analysed using the analysis module.

‘We tried to test successive improvements of the FMS by exposing each version to virtual operational conditions, by embedding real FMS in a virtual representation of an aircraft with high-representative crew to reproduce a flight between two airports with real meteorological data,’ continues Damour. ‘Thales has the ability to gather real meteorological data from anywhere in the world (open data) and incorporate them in the simulator. So they were able to reproduce virtually any flights within the virtual system.’ 

In terms of deliverables, two modules have been produced and sent to Thales: one is a simulation studio, which provides the capability to run massive simulation using a virtual pilot, and the second one is the data science module to run massive analysis – the culmination of all the analysis of the data for the validation.

‘The actual advantage now is to have a massive testing environment that can be used both for the FMS component in Large Passenger Aircraft and in Systems, so it really brings added value,’ reports Gurau.