Tired of Taking Your Shoes off at the Airport? This Technology Can Help

How machine learning can relieve the frustrations of airport security

Going through airport security is most travelers least favorite part of a vacation. It's bad enough having to walk through the scanner in stocking feet, but the inconsistencies between screenings may be even more frustrating. For instance, at one airport you may be told to remove all jewelry, while at another, keeping it on isnt a problem. Or perhaps you've experienced "liquid roulette" - a forgotten bottle of shampoo triggers an alarm at one screening point but is ignored in another.

 

Which raises the question, how effective are these screenings? You hear reports of airport screeners failing to detect firearms, but false alarms that lead to security guards rifling through your personal belongings happen more often that they should.

 

It's no wonder that passengers continue to be frustrated with the process. Meanwhile, airport security leaders are looking for ways to improve it.

 

Let the robots take a turn

 

That's where machine learning comes in. Machine learning is a branch of artificial intelligence that gets computers to act without being explicitly programmed. This technology allows computers to analyze real-world data and recognize patterns to make decisions or predictions. Self-driving cars are a form of machine learning, but this technology can be applied to everyday activities such as speech recognition, effective web search and, you guessed it, airport security screenings.

 

"It starts with collecting and analyzing the rich inputs received during screening", said Dr. Lisa Dolev, founder and CEO of Qylur, an intelligent systems technology company. Then, machine learning builds the ability to detect multiple classes of prohibited items and threats.

 

 

Screening technology that leverages machine learning can detect some items faster than human operators, which helps increase the efficiency of airport security lines. Machines are able to collect and analyze inputs from many sensors. Different inputs, such as multiple views of a bag at an airport and various energy levels of X-rays, are collected and analyzed in whats called "fused detection". This multi-dimensional perception enables machine learning to better characterize and identify problem items that are difficult for human screeners to detect.

 

"Machines can understand a huge amount of data much better than a human can", said Dr. Eric Xing, professor in Carnegie Mellon Universitys School of Computer Science. Once the machine collects that information, it will be able to determine patterns and identify which patterns need attention. The data that is gathered can then be used to better train humans.

Not ready for prime time - yet

 

But machine learning is not a perfect solution. If the machine makes a mistake, who will be liable? The blame for a machine error reaches a much wider network: the manufacturer, the software developer or the human responsible for running the machine. "Current artificial intelligence isnt quite ready to replace humans", said Xing. Neither the technology nor the legal concerns are in place just yet.

 

Perhaps the most effective way to use machine learning is to combine it with human interaction. In screening applications, the machine can do the initial screening to search for suspicious objects through the use of sensors and data, which streamlines the process for both customers and employees. If the machine catches something unusual, it can alert the human employee who can then conducts a secondary screening.

 

With machine learning, it is possible to do a very good job screening increasingly complex objects without needing to separate items. This lessens the hassle for travelers, increases efficiency and improves the overall experience of screening operations. With this technology, air travelers should never have to remove a shoe or bottle of shampoo again.

RELATED NEWS