Advances in artificial intelligence and machine learning (AI/ML) and the pursuit of open architecture standards for transportation security technology will improve the detection capabilities of existing security equipment while enabling these systems to evolve to meet new threats, says David Pekoske, the administrator of the Transportation Security Administration.
Pekoske also highlighted that the benefits of AI aren’t limited to detection equipment.
“Artificial intelligence is a promising horizontal emerging technology that promises to play a vital role, not only in our operational technology but our vetting systems, staffing programs and all aspects of agency operations,” he said on Jan. 31 during his State of Transportation Security address, hosted by the Government Technology and Services Coalition. “The potential impacts from AI on security of the homeland and upon our department’s operational activities, both positive and negative, make it imperative for DHS to take a proactive role in the use of AI systems and contribute to the national conversation on the secure use of this transformative technology. We will continue to proactively explore the future state of the passenger screening experience and where artificial intelligence will disrupt or improve this capability.”
Pekoske pointed to the computed tomography (CT) technology, which is relatively new to the checkpoint and is being used to screen carry-on bags for threat items, as the type of system that will benefit from AI/ML applications.
The checkpoint CT systems produce high quality three-dimensional images and generate large datasets from the bags that pass through them.
“CT and machine learning will allow TSA and its partners to develop dramatically improved detection algorithms based on these data sets,” Pekoske said. “For example, machine learning can help detect prohibited items like firearms, including components and potential 3-D printed weapons. Innovations in machine learning can potentially lead us to a state where we only review images on alarm, as we do in checked baggage screening. This will take some time but is an aspirational and very real future state.”
Pekoske also said that AI/ML and other technologies can improve the capabilities of millimeter wave-based Advanced Imaging Technology (AIT), which is used for on-person screening at airport security checkpoints.
“Coupled with emerging computer vision capabilities, AIT may continuously learn from millimeter wave data to better assess threats and prohibited items,” he said. “Computer vision and machine learning has the capacity to ingest data and improve algorithms in real time development.”
In 2020, airport operators and regulators worldwide published a paper supporting open architectures for airport security systems to improve interoperability, enable innovation and expand competition. The paper was endorsed by TSA.
“Internationally, TSA is establishing an open system modular architecture to advance our risk-based screening objectives, promote rapid response to evolving threats, and expedite the delivery of innovative capabilities to the front line,” Pekoske said.
Open architectures for security detection equipment would mean that third party vendors could provide applications, such a new algorithm to enhance threat detection, without a customer like TSA having to go through the original equipment manufacturer for the upgrade.
Pekoske said that open architectures will enable the integration of “emerging technology at speed,” and added that “New standards mean innovations and inventions seen in other sectors, often developed by lean and emerging startups can bring the same value to us.”