Nikolaos Vasiloglou IATA
Nikolaos Vasiloglou founded Analytics1305 and Ismion Inc., developed the PaperBoat machine-learning library and teaches at Georgia Tech.

Opinion: Disrupting Aviation With Artificial Intelligence

What can aviation learn from Silicon Valley to leverage the latest in AI?

Print headline: AI Disruption 


The International Air Transport Association hosted its annual Safety and Flight Operations conference in Montreal in April, where we were very happy to see small startups like SparkCognition, Safety Line and Dynamic Source getting attention with products that aim to optimize operations.

Artificial intelligence (AI) is ubiquitous, and inevitably it will disrupt the aviation industry—soon not just AI but other more traditional data science practices used by startups will be adopted by the slow-moving giants in aviation. There is a lot to be learned from agile and flexible Silicon Valley startups that can help us solve the difficult puzzle of progressing data fast without sacrificing safety and opening a war with unions and regulatory committees. Uber, Airbnb and others have managed to disrupt traditional markets while complying with regional policies and regulations.

IATA

Nikolaos Vasiloglou founded Analytics1305 and Ismion Inc., developed the PaperBoat machine-learning library and teaches at Georgia Tech.

The key factor in aviation is to use approved devices that do not attach to or interfere with the aircraft. As we pointed out at the Safety and Flight Operations conference, smartphones and high-definition cameras can open new avenues for collecting data that can be very useful for improving safety and customer experience.

Such devices will not substitute for the deterministic models used to operate the aircraft. They can, however, disrupt the industry by getting similar results once large amounts of data are analyzed. These results can be used as another highly dependable source of information in decision making.

MULTIPLYING THE EYES IN THE COCKPIT

One of the greatest breakthroughs of AI is in computer vision. Smart cameras that can run deep learning models (the basic ingredient of AI) can scan what is happening in the visual field of pilots. One of the biggest complaints is about obstacles encountered during takeoff that often are not recorded in databases. An Amazon DeepLens or a GoPro camera can easily collect that type of information. Cameras can also capture an instrument reading and process it real time. Although the flight recorder does the same job, real-time access to it is still far in the future.

EVERY CELLPHONE CAN TELL A STORY

Smartphones are one of the most inexpensive and common sensors of our time. Cellphones can collect information about sudden accelerations and audible noises coming from vibrations. They can also measure unusual activity in the cabin that comes from temperature discomfort or low cabin pressure. Moreover, smartphones can give a good estimate of the height and weight of a passenger that can lead to better balancing of the airplane. Keep on brainstorming and soon you will see that there is a ton of useful data you can collect and mine.

WHO OWNS THE DATA?

One of the most common counterarguments in our vision is ownership of data. First of all, are passengers willing to share their personal data? I will be very positive and say “yes.” Airlines have no motivation to sell this data or use it to annoy customers with ads. Besides, some incentives with frequent-flier miles can bend any hesitation. But again, airlines are not like Facebook—all they have to care about is adopting proper cybersecurity practices so they don’t leak information.

Internet service providers like GoGo, GEE and ViaSat can also be part of the game. They have the infrastructure to collect data and transmit it to data centers on the ground. Offer some free bandwidth to the users, and they can become owners of the data, which they can sell to airlines. 

Second, are pilots willing to allow cameras in the cockpit? Let’s give the pilots what they want: privacy. Let them be the owners of the data and have it in their discretion to share the data. They are fully entitled to review any footage before they share it with the AI company that will do the processing. Contracts like that are very common in the internet industry.

BEING MORE APPEALING THAN GOOGLE

While everybody is struggling to find AI talent, Apple, Google, Microsoft, Amazon and Facebook are sweeping the market by offering free food, scooters and other perks. Airlines may not be considered as fun a place for a computer science graduate to work, mainly because of corporate bureaucracy and boring projects. But if you bring the Silicon Valley fun stuff and give them the opportunity to tune their ideas and try them on the fly, then you have won them. An airline wants to hire computer science geeks because  they are the seeds of innovation.

A SEA OF APPS

Modern AI can bring a fresh air of innovation to the industry. It can enrich the data sources that are currently available and break any monopolies on data. Then a sea of apps can be built on top of them, without interfering with regulators and other inflexible structures. 

Nikolaos Vasiloglou founded Analytics1305 and Ismion Inc., developed the PaperBoat machine-learning library and teaches at Georgia Tech.

The views expressed are not necessarily shared by Aviation Week.

 

TAGS: Technology
Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish