Pangiam in July expects to begin the first of five aviation security pilots this year that integrates its prohibited item detection algorithm with a checkpoint computed tomography (CT) scanner that detects explosives in carry-on bags, marking the first use by an airport of a third-party algorithm with the checkpoint scanners.

The initial pilot will occur at Glasgow airport in the United Kingdom in partnership with Google Cloud and AGS Airport Ltd., the airport operator. Pangiam expects additional evaluations to begin later this summer at two major European airport hubs and also at two airports elsewhere in the world, Alexis Long, the company’s chief strategy officer, told

HSR in an interview in late May.

Pangiam’s algorithm will be integrated with the DETECT 1000 checkpoint CT system, which is being supplied to AGS by Integrated Defense & Security Solutions (IDSS) under a recent contract to conduct trials at three U.K. airports, including Glasgow. IDSS is based in Massachusetts and has been competing to supply the U.S. Transportation Security Administration with checkpoint CT systems.

Long said that the choice to go with IDSS was based on a decision that company had already made to pursue an open architecture approach to allow its system to interoperate with third parties and stream images in an open format.

Pangiam’s image processing capabilities and Google’s [GOOG] artificial intelligence and machine learning (AI/ML) computer vision tools are combined to allow Google’s technology to read the CT images to detect prohibited items. Pangiam also has ML models for security detection that help Google create ML models to train the AI technology to identify threats within carry-on bags.

Initially, the Pangiam solution will be connected to the DETECT 100 but run in the background in “shadow mode” to learn what’s passing through the scanner to get “used to the normal pattern of life data it’s seeing,” Long said. Once the technology is dialed in it will go live and when prohibited items such as guns and knives are present the algorithm will highlight for the operator where in a bag a potential threat may be hidden, he said.

All along, the CT system will still be detecting for potential explosives hidden in bags.

“So, if you look at the concept of operations with IDSS as an example, the DETECT 1000 is going to find the explosives and we’re going to find the weapons and together the screen readers will be able to just focus on those bags that have got the greatest chance of having a threat item in them,” Long said.

The opportunity to begin proving out open architectures in checkpoint technology, in this case the CT systems, potentially offers a game-changing paradigm for aviation security regulators, airport operators, and other stakeholders in that they will be less reliant on the original equipment manufacturer to upgrade equipment. In this case, Pangiam and Google as a third party, bring a new detection capability that could just be plugged-in to the CT system, enhancing the system and even increasing competition among vendors for new detection capabilities and other improvements.

David Pekoske, the head of the U.S. TSA, is a big proponent of being able to upgrade traditional aviation security technology, in particular the checkpoint CT systems, with third-party software to enhance threat detection.

As the pilots expand to other airports, Long said they expect to add at least one more checkpoint CT vendor to the mix will also bring open architecture to the mix. So far, TSA has purchased checkpoint CT systems from Smiths Detection and Analogic, with the latter the only company certified to the agency’s latest explosives detection standards as of earlier this year. Leidos [LDOS] is also competing to provide the checkpoint CT systems to TSA and globally.

Once more airports begin participating in the pilots, Pangiam plans to evaluate additional capabilities, including aggregating data between different manufacturer’s CT systems that are located at different airports.

Long described two reasons for this. One is that it brings “national situational awareness” to different countries, he said.

For example, similar to the complex threat posed by the 9/11 hijackers that boarded separate flights, there could be a “scenario where the same homemade knife is passing through three checkpoints on flights going to New York,” Long said. The technology becomes an “alertness tool to potentially spot patterns of stuff and we know aviation gets targeted by complex threats.”

Another reason for being able to quickly aggregate data between airports and different checkpoint CT systems to be able to identify new threats, and then rapidly train the model against threats seen elsewhere, Long said.

“So, you could see a new concealment method, or indeed, not a threat but a new false alarm at an airport and that model will be able to very quickly disseminate to other airport to share learning,” he said. “That’s really where the power of the Google Cloud comes in, because it allows us to aggregate that at that kind of planet scale where we can quite rapidly shift learnings from one region to another.”

Pangiam calls its collaborative effort with Google Cloud Project DARTMOUTH.

European Prohibited Item Detection Standard

Long also said that Europe has developed a standard for automatically detecting prohibited items in carry-on bags. This fall, the European Civil Aviation Conference, which certifies security technologies to various standards, is expected to begin testing the new standards for automated prohibited item detection, he said.

“And if they commence testing this autumn as we suspect, then effectively the market has been created because there is now a regulatory path to be able to use these technologies, which is quite exciting,” Long said. “So, in practice, what it will mean is we can start using our technology to speed up screen reading by directing officers to certain parts of a bag that are most likely to contain a prohibited time. And so, basically airports will start realizing operational gains.”