Low Cost & Regional

How artificial intelligence is now supporting the aviation industry

artificial intelligence
photo_camera EasyJet stakeholder event showcasing the new Airbus a320Neo Aircraft at Bristol Airport.18/01/2018Photo by Simon Galloway Contact photographer: Simon Galloway 07810 638162 [email protected] www.simongallowayphotography.co.uk

Aviation has frequently been accused of lagging behind other industries when it comes to artificial intelligence. Stephanie Taylor rounds up the ways in which airlines are catching on.

Airlines and airports are now embracing new technologies and turning to artificial intelligence (AI) to support their customer service.

Over the next three years 52 per cent of airlines plan major AI programmes or R&D, and 45 per cent of airports will invest in R&D in the next five years, according to the SITA 2017 Air Transport IT Trends Insights.

Airlines are looking at how technology can help minimise the impact of disruption on the passenger experience and their business. Over the next three years, the SITA report says 80 per cent of them plan to invest in major programmes or R&D into prediction and warning systems, which rely heavily on AI.

The International Air Transport Association (IATA) and Airports Council International (ACI) have noticed AI’s value too.

In October 2017, they came together to launch the New Experience in Travel and Technology (NEXTT) initiative, aiming to optimise the use of emerging technologies in the face of growing passenger numbers. AI is highlighted as a priority, specifically in relation to its ability to improve real-time decision-making and, therefore, efficiency.

As detailed in a 2017 report titled ‘The Future is Predictable’, SITA itself is using AI to tackle flight disruptions, which the company says can cost the air transport industry $25 billion each year.

Since 2016, the SITA Lab has been evaluating disruption detection and prediction capabilities. To do so, it has collected a massive amount of existing data on flight movements, weather, and flight delays, but has also begun mining three new data sources: Notice to Airmen (NOTAMs), news feeds and Twitter feeds.

There was much work involved. SITA decoded information from the NOTAMs using the capabilities of IBM Watson – a question answering computer system capable of answering questions posed in natural language – which it trained to understand air transport specific acronyms.

Barnett says LCCs needn’t worry about how the system will affect their bottom line

For news feeds, the company employed natural language processing to find publicly available news containing information that could affect air transport stakeholders’ operations. Twitter is being used to flag problems reported by passengers at airports.

These data streams are now being harnessed by SITA to create short-term (under 48 hours) and long-term (48 to 72 hour) delay prediction platforms. Eventually, a delay predictions algorithm and disruption warning feed will be incorporated into SITA’s services to the air transport industry.

Customer service is one of the most important facets in air travel, but famously the need for it begins when passengers start thinking about booking flights, not just when they get to the airport. This is why chatbots, which are based on AI and machine learning, are becoming prevalent.

Indeed, SITA’s aforementioned IT trends report predicts that 68 per cent of airlines and 42 per cent of airports plan to adopt AI-driven chatbot services by 2020.

Mirabeau, the digital agency headquartered in Amsterdam, helped create Transavia Flight Search on Facebook Messenger, and it pronounced in its case study of the project that the reason for such expansion in this area is the growing use of mobile through applications like WhatsApp.

The company stated, “chat is quickly becoming the interaction of choice among mobile users.”

Transavia was won over by a prototype Mirabeau created for them using Microsoft Azure’s Cortana. Now live, the chatbot guides potential customers through the ticket selection and purchasing process using a mix of text chat and click-based automation.

“The key was finding the points in the interaction where chat can provide real added value,” claimed Mirabeau. “Then, we combined that element with some simple click-based functions to accelerate the process.”

Machine learning means the Transavia Flight Search tool is growing smarter with every transaction, and the Mirabeau team can make “improvements based on real-time feedback and analytics.” Edgard Beckand, Mirabeau’s strategic director, asserted, “We’re looking to expand the experience to seat selection, managing payments and providing excellent customer service.”

Mexican LCC Viva Aerobus is taking off with a customer intelligence cloud system

For Transavia’s managing director, Mattijs ten Brink, the move was about proving the airline’s philosophy – ‘we need to be where our customers expect us to be’ – but for Mirabeau, it appears chatbots are just the beginning.

Through a partnership with Microsoft, the company has access to its full suite of cognitive services, which include not only natural language understanding, but also facial recognition and emotion detection.

Talking about the ‘conversational interfaces’ enabled by this technology, Beckand said, “We expect this technology to have a bigger impact than the smartphone revolution.”

But AI isn’t particularly new. Low cost carrier (LCC) easyJet said it first began using AI solutions in 2010, and in 2015 introduced its first head of data science, Alberto‎ Rey-Villaverde, to accelerate the carrier’s use of artificial intelligence.

Soon afterwards, it made headlines for using AI to predict how many items of certain food and drink were required on different flights, improving both the passenger experience as well as minimising waste.

In contrast to most legacy airlines, which struggle with multiple silos of data, Villaverde acknowledged when he was appointed to the role that easyJet did have an advantage when it came to AI: “All bookings are made through easyJet.com, we fly one aircraft type and we only fly short-haul.

“This combination of simplicity and scale produces an enormous amount of data. AI can make sense of this huge volume of data and make it work for us.”

Boxever, a customer intelligence cloud,  is helping airlines with a more complex setup to integrate disparate data to ensure personalised offers and ultimately improve conversions. In a case study of its work with Mexican LCC Viva Aerobus, Boxever said it used AI to act as VivaAerobus’ ‘brain’.

The company pulled data from the airline’s website, Navitaire reservation system, call centre and mobile site to create a single view of each customer and communicate with them in real time on the most appropriate channel.

One of the solutions deployed by Boxever catered specifically to the Mexican market. Regarding cash payment, Aurelius Noell, director of eCommerce and commercial IT, VivaAerobus, testified, “This payment option is huge in the Mexican market; it accounts for 30 per cent of our transactions.

artificial Intelligence: FliteTrak Viator Aero
FliteTrak Viator Aero – sensors, data loggers and tablet showing app display

“With Boxever, we are able to automatically send cash payment customers an email that reminds them to pay in 24 hours to avoid cancellation, shows their nearest store, and also provides other payment options if they would like to change.

“It’s a win-win for everyone. Our airline receives cash payment for more bookings and customers don’t have to do any work to find out where the store is, or they can easily switch payment to credit card if they are not near a store,” Noell concluded.

If Boxever’s figures are to be believed, the benefits of harnessing AI are huge. Supposedly, within one week of deploying six Boxever applications, VivaAerobus accrued nearly $1 million in incremental revenue based on one week’s performance and went on to increase its Net Promoter Score by a staggering 60 per cent.

It seems the ability of AI to limit passenger pain points is endless. Although not covered in its case study with VivaAerobus, Boxever also believes airlines can leverage AI along with predictive analytics to prevent overbooking issues, which have been the source of much bad press.

AI isn’t just limited to pre-flight interactions either. New companies are coming up with ways to use it onboard. In ‘The Future is Predictable’, SITA states 17 per cent of airlines are implementing ‘Internet Of Things monitoring capabilities’ focused on passenger seating to gauge factors including a traveller’s temperature or tiredness.

UK-based FliteTrak has come up with its own intelligent remote monitoring technology, ViatorAero, to do just that. It works by using sensors installed in the seat alongside FliteTrak’s proprietary SpriteAero boxes (one of which can handle the data for up to four seats).

“We have carried out tests with two to four strain sensors per seat – two is enough to detect movement and agitation. Temperature, humidity and air quality monitoring is built into the SpriteAero box itself,” explains Andrew Barnett, joint managing director of FliteTrak.

ViatorAero isn’t just about passenger comfort – it can take safety into account by providing information on seatbelt closure and mobile phone activation.

To enable predictive maintenance, Barnett says “SpriteAero can be provided with built-in screen, which can display seat fault data. The ViatorAero hub can also be fitted with a colour screen, again useful for test and diagnostics purposes.”

artificial Intelligence helps Transavia connect with passengers
AI is changing how Transavia connects with its passengers

The system covers overhead bins too, a well-known pain-point for both passengers and crew on busy flights. “We are able to activate locking mechanisms within overhead bins, and this adaptation of the same SpriteAero box can detect light and proximity, so we could determine available space,” Barnett elucidates.

The SpriteAero boxes collect their own data and the data from the in-seat sensors and send to crew tablets or onboard IT systems in real time using a near live data transmission (NLDT) algorithm, which “hunts for and uses the best data channels to deliver data from sensors as quickly and reliably as possible,” says Barnett.

The ViatorAero system has been designed so LCCs needn’t worry about how it will affect their bottom line: “SpriteAero units are very lightweight, draw microamps and can run from rechargeable integral batteries or low-power aircraft DC voltages.” Interestingly, Barnett continues, “We are testing passenger movement or aircraft vibration as a method of powering sensors.”

In terms of installation times, Barnett estimates that it would take five minutes per SpriteAero box and, if the seats had removable seat pads, installing the sensors would take minutes each too.

Worried it’s all a bit Big Brother? Barnett notes that the system only stores and transmits information about a certain seat number, meaning the data is anonymous.

However, the capabilities of ViatorAero raise the question of whether it would be beneficial for the airline to build a profile on a passenger so that if you’re consistently a restless flyer, for example, the crew know to give you some extra attention. It’s predictive maintenance, but for people.

The ViatorAero system has been shortlisted for this year’s Crystal Cabin Awards, and the company plans to keep on innovating.

It is continuing to work with AI at its Innovation Hub at Daedalus Park in Hampshire in partnership with Microsoft’s machine learning systems and is currently in the process of equipping the facility with half of an aircraft interior for experimentation purposes.

Jim Peters, chief information officer for SITA, noted in a recent blog that AI complements the movement of passengers towards self-service.

With airlines always talking about the importance of the ‘human touch’, it’s lucky AI is making technology increasingly human and giving airline staff the time to provide customers with increasingly meaningful interactions.

Editor’s Note: The post was originally published in April 2018.

Sign In

Lost your password?