Feature Article: How an S&T Industry Day is Shaping the Future of Security Screening

One of the Science and Technology Directorate's (S&T) key mission priorities is to promote lawful travel and trade across air, land, and sea via our ports of entry. A new Cooperative Research and Development Agreement (CRADA), currently in development, will enhance how we work with industry and how we leverage the power of artificial intelligence (AI) and machine learning (ML) to do just that.

S&T is at the forefront of research and development that is driving the next generation of automated, less intrusive, cybersecure, and cost-effective aviation screening solutions. By enhancing security screening technologies for people, cargo, baggage, and goods at airports and border checkpoints across the country, S&T is directly supporting the Department of Homeland Security's strategic priorities to modernize and secure the homeland.

Our Transportation Security Laboratory (TSL) tests, evaluates, and certifies new technologies and algorithms for nationwide deployment. These efforts rely on high-quality data to transition new technologies from the lab to operational environments. Advancing the security screening landscape requires innovative approaches and robust public-private collaboration, which is why S&T and TSL are establishing the CRADA to encourage the private sector to join forces with us.

Priority: Building Compatible, Open Architecture Security Systems

S&T is prioritizing development of a secure system for data sharing among approved industry partners to enable the Transportation Security Administration (TSA) to seamlessly deploy multiple algorithms across different types of screening equipment.

The Screening System Data Sharing Consortium CRADA will allow approved companies to collect, validate, annotate, curate, synthesize, and distribute screening system data to authorized software developers, enabling them to produce robust and reliable threat detection algorithms for Transportation Screening Equipment (TSE). The CRADA will also allow all algorithms produced by software developers to be utilized on all U.S. screening equipment, supporting interoperability and innovation.

TSA is currently developing a cloud-based data repository, the RCA Data Transfer Hub, to house all data collected by screening equipment and algorithm developers. The data consortium is envisioned as a collaborative marketplace connecting algorithm developers, equipment manufacturers, software developers, testing laboratories, synthetic data developers, and front-end users. This will foster innovation and ensure screening solutions meet the needs of more than 450 domestic commercial airports and other screening technology users.

The emergence of ML-enabled algorithms makes the need for such a data consortium even more critical. While the performance potential of these algorithms is widely recognized, TSE vendors struggle to obtain sufficient and diverse data to reliably train their models. The Consortium will enable members to pool large volumes of diverse data-accelerating training, reducing costs, and enhancing the robustness of the resulting algorithms. TSL intends to apply targeted guidelines regarding the type of data shared with third parties, ensuring compliance with classification standards for restricted materials.

"The advantage of bringing industry together is that this problem is too big for one organization to solve alone. TSL doesn't have the resources, and individual industry members can't collect and curate the amount of data they need to train and test their algorithms," said TSL Director Dr. Christopher Smith. "Together, we can manage that process in a way that expedites the delivery of validated technologies to our TSA customer."

A great deal of thought and feedback went into the development of the CRADA. To ensure it meets the needs of all stakeholders, TSL hosted an industry day earlier this year to gather input that will ultimately help transform the screening landscape.

Priority: Convening Government and Industry to Share Aviation Security Data More Effectively

The two-day event drew companies from a variety of sectors, including AI/ML, synthetic data, and security screening equipment manufacturing. Together, they discussed efforts to move towards open architecture and a new approach to evaluate third-party algorithms that can be integrated into screening equipment.

Industry representatives also discussed how to efficiently share large amounts of data, use AI in daily operations, and apply computer-generated (synthetic) data to train algorithms. Bringing together the people and technologies involved in keeping air travel and borders safe led to productive discussions about how industry and government can collaborate to increase automation and reduce the time it takes to develop a TSA-certified screening system.

Priority: Creating a Collaborative Space for Research, Development, Test, and Evaluation

One of the main goals of the industry day was to develop a framework for a data consortium-an agreement among companies that create algorithms, manufacturers of security screening equipment, data-storage organizations, and the government for how to work together and share data responsibly. Once in place, this consortium will enable members to securely share data in one place, making it easier to support research, testing, and validation of new screening technologies. This shared data hub will help members train and improve their equipment faster, speeding up the development, testing, and deployment of new screening technologies at airports, borders, and event ports of entry. This large-scale collaboration will help advance the next generation of screening equipment.

The event highlighted the substantial need for vast and comprehensive datasets to train ML-enabled algorithms that can detect potential threats. Because manually collecting the needed training data is impractical, synthetic data was recognized as a vital resource to train not only screening technology, but Transportation Security Officers as well. During the discussions technical experts in physics, engineering, and computer vision proposed a new process to verify and validate screening technology, ensuring that synthetic data can reliably replicate real-world conditions.

Feedback captured helped clarify how the government intends to structure the consortium moving forward, including developing a practical charter, rules for membership, and operational methods for securely storing, transmitting, and accessing collected datasets.

"It takes a community to drive innovation, which is why S&T and TSL are grateful for all of our industry partners who are taking the next step in advancing the transportation security screening landscape," Smith said.

The TSL industry day showcased the power of S&T's strong partnerships to modernize security screening and demonstrated the Directorate's, and the Department's, commitment to leveraging cutting-edge technologies to strengthen homeland security.

S&T will share announcements when the CRADA is posted on SAM.gov in coming months. For related media inquiries, contact [email protected].