Improving airport security with AI technology

Alexis Long (AL): Ever since around 2018, we recognised both in the US and across many European partners that the pace of innovation we were seeing in aviation security wasn’t really keeping up with what we wanted. 

We took a number of different actions to try and stimulate innovation more. That included things like putting money into venture capital funds out in Silicon Valley. We created new ways for companies to trial technologies and so on. 

One of the really important things we did was to try and break what we thought was a barrier to new entrants entering the market, which was almost a stronghold on data. That meant that it was very difficult for new players in this space to get hold of the data in a common format, at a pace that they need it to be able to do creative things with it. 

Airports Council International (ACI) spearheaded a campaign that got about 25 of the world’s largest buyers of aviation security together and said: “From now on, when we buy new technology, we’re going to want certain things. We’re going to want that data open, so that other people can bring new technologies to that data. We want the ownership rights sorted, and we want the cybersecurity rules sorted.” 

What that document really did was break open the market to new entrants, and what we’re seeing now is airports across the world taking that document and embedding it in their procurement framework. Every time they buy a new kit, they have all the rights and ownership access to the data.

AL: When we began looking at that data across lanes and checkpoints at airports, we started being able to look at other patterns that could indicate coordinated threats. That’s super important, because what you see if you ever track terror plots by sector on a graph, you often see aviation tends to get the most complex ones. It’s rarely individuals that decide “I’m going to pick up a knife and come to the airport today and board a plane”. 

The plots that seem to be disrupted are often a lot more complex. What we can do by digitising all of this information is start looking at things, such as detecting component parts of a firearm and starting to say, “we’ve now found something here that in itself isn’t a threat, but when you look at it in conjunction with those others it is”. 

We’re really excited about this. We are partnering with AGS airports to rapidly accelerate the development of this technology. As airports build back, there’s a real desire for them to make sure that their important human resources are spent where it’s most needed.

AL: The thing about software is it doesn’t change performance over time. It doesn’t get distracted, doesn’t get tired, it doesn’t get overworked. One of the key things is that we can bring consistency to that checkpoint operations. 

You might go through lane A and have an item removed and not at lane B next time you travel. That’s really important for airport operators, because consistency is something that’s very hard for them to manage in terms of demand and flow to the airport. It’s good for the passenger as well, because you can start being better assured about what you have. 

For the security officers, what we want to do is basically say to them: “You don’t have to worry about this set of bags. You concentrate your human effort on resolving these ones where we think there’s something wrong with it. It now requires a human interaction to determine whether that’s a deliberate attempt to evade security or accidental.”  

There’s a large opportunity there for officers to focus on where they’re most needed. 

From a security perspective, we can start putting things together and stitch together lanes and different checkpoints. We can start bringing automation to things that are traditionally very difficult to do like human intuition aspects.

Ha McNeill (HM): When you look at the airport environment, you’re seeing year-on-year growth of passengers at about 5%. This is a pre-Covid-19 figure, but we anticipate that will come back. 

You’re already seeing challenges with processing that amount of people through a checkpoint, and airports are historically limited when it comes to infrastructure, so unless you drastically expand and build a new airport, you’re dealing with the same footprint. So how do you move that additional volume through and still give people that better experience? 

Bringing the power of AI and machine learning really does help streamline some of the security processes and move passengers through the checkpoint. It allows agencies to tackle that volume issue.  

AI brings continual improvement to the checkpoint. It is continually learning and improving. You don’t really have to do a whole lot in the background. 

AL: We’re working at group level at the moment. We’re really working on two things. We are doing a series of offline trials and pilots, where we have a first version of the software already developed. 

We’re looking at various components of that, such as the look and feel, the user interface, and what an appropriate performance rate is. Then by Q2 this year we’ll be moving to live pilots, deploying this technology on the lanes. To begin with, that will be in what we call ‘Shadow Mode’. It will be running on the lanes but not impacting an officer’s decision making. 

We’ll be using that to refine our settings and ensure performance is where we want it to be. Then we have a roadmap forward for how we can dial up to the amount of computational support, and we’ve focused security officer attention and that will come throughout this calendar year. 

It’s a very aggressive pilot in terms of timetables. This isn’t something that’s going to sit as a one or two-year research project. As airports buy and procure this technology, we want it to be ready to be turned on – so there’s quite a large push to move that forward quite quickly.

AL: Absolutely. I think there’s a lot of things that need to be set up to make sure that along with us, other people can follow – because we want to create an ecosystem of suppliers where software companies big or small can all participate. 

We’re pushing forward to create that environment now, but I don’t see anyone going back after this. I think the requirement to make sure that officers have as much support as possible in their decision making and can focus on fewer bags, or items that pose the greatest risk of harm, is something that’s here to stay. 

HM: I agree. I think for all of the pressures that the industry and the regulators are currently facing, we wouldn’t see this going away and are viewing it as creating the potential to take this technology and apply it to different use cases within the aviation industry.