AWS for Industries
In the News: Today’s in-flight experience is brought to you by machine learning
The below article originally appeared in WIRED. 
Inflight entertainment in the pre-internet era was limited to the books and magazines that could be crammed into a passenger’s carry-on—or bought last-minute at the airport. But today, from the array of entertainment options to inflight Wi-Fi, technology has made today’s flying experience markedly more enjoyable.
Now, Panasonic Avionics is further modernizing air travel by personalizing the inflight experience for passengers. “Unfortunately, most people still dread air travel,” says Samir Lad, Panasonic Avionics’ head of digital architecture. “But we’re hoping to change this fundamentally using modern cloud and machine learning technology offered by AWS.”
Using advanced machine learning and deep learning algorithms on AWS, Panasonic Avionics built NEXT Cloud, a commercial aviation platform for building a modern digital cabin. NEXT Cloud anonymously collects inflight data about passengers—such as the kind of films they like to watch, the food they like to eat and the quality of their Wi-Fi—to customize flight offerings for passenger tastes. And in turn, these insights help drive operational efficiencies for the more than 300 airlines Panasonic Avionics works with.
Adding a Personal Touch to Traveling
Research suggests that airline passengers are open to personalization and even expect airlines to anticipate their needs. Singapore Airlines is one of Panasonic Avionics’ partners leveraging passenger data to create a more consistent experience throughout a customer’s entire flight journey. The two companies collaborated to create a mobile app, myKrisWorld, that acts as a personalized portal for passengers at every stage of their travel: from pre-flight booking, mobile check-in, and luggage tracking. They can also customize the portal to a specific language, bookmark favorite content and be served recommended content based on their browsing and viewing history.
Panasonic Avionics and Singapore Airlines were recognized with the Best Personalization Innovation award at the 2018 Airline Passenger Experience Association Awards for building the myKrisWorld platform.
Data-Driven Entertainment at 31,000 Feet
Beyond customizing the experience for individual passengers, Panasonic uses machine learning to help its partners understand passenger data on a macro scale. Take entertainment, for instance. Panasonic’s data (which, Lad stresses, is anonymized) suggests most airlines are falling short of delivering the types of media passengers actually want—and it’s using these findings to help airlines make smarter decisions about what they offer in flight.
Data gathered on the NEXT Cloud platform revealed, for example, that airline passengers don’t really care to watch the latest blockbuster when they’re in the air; they’d rather re-watch a classic. Lad says they can only guess the reasoning behind trends like this one:
“If there is a new movie and my family is waiting for me to come back from traveling and watch it in the theater, I don’t want to spoil the fun by saying, ‘Hey, I already watched that movie on the plane,’” he says.
Flight duration, route, and seasonality also impact these broad preferences. Panasonic’s data shows passengers coming to and from Asia are more likely to spend their flight watching game shows, for instance. Passengers listen to pop music on USA to Asia routes. And on flights to and from certain southeast asian destinations, passengers are more likely to pass the time playing video games on their IFE systems.
Armed with these insights, Panasonic’s airline partners can adjust what content to offer. Not only does this help the airlines provide a better experience for their passengers, but it also avoids the cost of licensing fees for content that no one watches.
Toward a Connected Travel Experience
Efficiencies like these are powered by AWS’ machine learning solutions—right from organizing and storing the data to analyzing and making recommendations around optimization. The same concept extends to identify which food items are popular and which often go uneaten, helping airlines cut down on storage costs. It can also determine the optimal location for Wi-Fi hotspots and antennas in an aircraft, increasing the quality of coverage—which, Lad says, is a major pain point for many airlines today in ensuring customer satisfaction.
Panasonic is currently looking to expand the application of machine learning into other areas of the passenger experience, like shopping. As with food and movie choices, its technology can help airlines identify products they should stock on certain flights—particularly based on destination: When flying into rainy climates, airlines might stock more umbrellas. For flights to sunny places, it can help passengers buy sunglasses, bathing suits and other items travelers often forget to pack.
Many flights also need to know how much duty-free merchandise to stock—like luxury perfumes and designer handbags. But the technology can also help move that inventory by using data-based insights to decide what discounts to give to passengers.
“Marketing teams on the ground can activate and monitor these campaigns in near real-time, and move over-stocked inventory,” Samir Lad says, “All of that becomes possible because of this hyper-connected IoT infrastructure, massive compute power, and the insights that we can derive using machine learning.”
Beyond inflight retail, Lad says, he sees a future where airlines can extend this tailored experience even outside of the airport terminals—helping travelers make dinner reservations or book an attraction at their destination.
“We’re barely scratching the surface right now,” Samir Lad says, “And the sky is the limit.”
___________