In December, Pangiam and Google Cloud announced a partnership aimed at doing what so far has seemed impossible in the security detection business, which is coming in as a third party to bring advanced technologies to enhance the detection equipment manufactured by other companies, whether working through the original equipment manufacturers or their end customers such as the Transportation Security Administration.
Project DARTMOUTH, the name of the collaborative effort between Pangiam and Google’s [GOOG] cloud business, combines Google Cloud’s artificial intelligence (AI) and machine learning (ML) computer vision tools with Pangiam’s image processing capabilities to allow Google’s technology to read the computed tomography (CT) images produced by checkpoint CT systems to detect prohibited items.
Pangiam also brings to the table ML models for security detection to help Google create ML models to train the AI technology to identify threats within carry-on bags, Alexis Long, Pangiam’s chief strategy officer, told HSR in an interview earlier this month. In addition to detecting various threats, the software can provide alerts to users, he said.
Long also said that updates to the software can easily be pushed to the edge devices as often as needed.
The AI/ML technology, which works with any security detection equipment at the edge, can connect all the screening equipment at an airport or airports, enabling aggregated threat detection across sensor types, checkpoint lanes and checkpoints, Long said. The technology can also scale across different airports, he said.
Aggregated threat detection refers to being able to spot coordinated threats, such as disassembled weapons, across different bags, lanes and checkpoints, Pangiam says on its website. The technology can also enable the detection of nationally coordinated threats, it says.
The announcement in December said that the technology will be tested at the security facilities of AGS Airport Ltd., which owns and operates Aberdeen, Glasgow and Southampton Airports in the United Kingdom. Offline testing began last year and operational testing with checkpoint CT systems will begin in the first quarter of 2022, Long said.
The testing will also look at aggregating threat data across the airports, he said.
Checkpoint CT systems are being deployed at airports worldwide, representing a significant advancement in scanning carry-on bags for potential threats because of their ability to automatically detect potential threats, particularly explosives. The systems also allow travelers to leave their personal electronic devices and laptop computers in their bags, adding a convenience factor at checkpoints.
The technology being advanced with Pangiam and Google Cloud will improve the existing technology on manufacturers’ systems by providing real-time detection of prohibited items, spotting anomalies such as items that aren’t normally carried by passengers, and even differences in an electronic device from a thousand other similar devices and providing an alert for further examination, Long said.
“We interrogate data for patterns,” Long said.
Pangiam describes the AI/ML technology being developed and applied in Project DARTMOUTH as akin to “human intuition,” which is being able to figure out when something is “not quite right,” even though some items aren’t on a prohibited list.
“But, because terrorist groups are constantly trying to develop ways of concealing threat objects, it will mean there are always some objects that aren’t yet on those lists,” Pangiam says in a YouTube presentation about Project DARTMOUTH. “So, officers use their intuition to alert them to something when it is not quite right. Their intuition can stop new and dangerous items from being taken onto an aircraft.”
OEMS have largely developed proprietary systems so that don’t easily integrate with third party apps due to a lack of open systems architecture, which has made it impossible for end customers to acquire software and other technology separately and then install it on the security detection equipment that they previously purchased.
Long said that Project DARTMOUTH complements existing aviation security technologies, adding that “we are working with some” of the OEMs. Ultimately, it’s up to the end customers to “structure this but we want to work with everyone,” he said.
There is a new push by end customers for an open systems approach to security detection equipment, prominently highlighted nearly two years ago in a paper published by airport authorities in London and Oslo and endorsed by airport operators and regulators worldwide, including TSA, for open architecture approaches to airport security equipment.
Open systems approaches are increasingly being demanded by buyers of security detection equipment, Long said. While there was resistance in the past by manufacturers to open systems approaches, AI/ML technology has significantly advanced in the past five to 10 years and having access to data from different manufacturers’ machines is less of a problem now, he said.
Pangiam is also interested in pursuing other markets such as customs and trade, and public safety to apply its AI/ML technology, Long said.