Customs and Border Protection (CBP) this week will host a virtual industry day to discuss its upcoming procurement of algorithms that can automatically detect anomalies inside cargo conveyances and passenger vehicles scanned by X-Ray imaging systems when entering the U.S.
The agency wants open-source solutions that leverage artificial intelligence and machine learning (AI/ML) that speed processing times for scans, improve security, and reduce manual requirements all without restricting the legitimate flow of travel and commerce.
The non-intrusive inspection (NII) anomaly detection algorithms (ADA) industry day is scheduled for Jan. 26. CBP made the announcement on Jan. 13 and also released a draft statement of work for the ADA.
The initial goal of the ADA will be to help operators of NII equipment to detect anomalies and contraband found in images captured by the scanning systems and the ultimate goal is for the algorithms to automate image analysis for passenger vehicles and cargo conveyances.
CBP wants open-source algorithms to avoid being tied to only algorithms supplied by respective NII equipment providers. Ultimately, the agency wants the algorithms to automate image analysis.
“The desired end state for the NII Systems Program is an open-source platform that supports autonomous image analysis for low-risk trade and travel, provides assistance to CBP officers for review of high-risk trade and travel images, and the analysis of more complex images,” says the draft statement of work. “Therefore, it is critical that all algorithms have the ability to be integrated easily into the larger NII platform and not solely be integrated into closed proprietary applications/systems.”
CBP in fiscal year 2023 plans to begin to significantly ramp up deployments of new NII systems, low-energy portals (LEPs) to scan passenger occupied vehicles, and multi-energy portals (MEPs) to scan the occupied cab of a truck at a low-energy and the cargo conveyance at a higher energy. These systems will be deployed at pre-primary inspection lanes at land border ports of entry along the southwest border, meaning a greater percentage of vehicles overall will be scanned by NII equipment before entering the U.S.
However, the NII systems require manual image analysis to sort find anomalies, such as packages hidden inside a vehicle compartment or items not included on a cargo manifest. The ADA algorithms would help operators more easily and quickly detect these anomalies.
Homeland Security Secretary Alejandro Mayorkas earlier this month said that DHS has an upcoming business opportunity for AI/ML integrated with CBP’s NII systems. Congress in the recent appropriations bill directed CBP to get moving on AI/ML for the imaging technology.
Technology demonstrations deployment of the LEPs and MEPs has generated “a large volume” of images as vehicles and cargo enter the U.S., data that CBP is collecting in its cloud computers and will be available for analysis, the agency says.
The forthcoming business opportunity follows recent prompting by congressional appropriators directing that CBP move ahead on introducing AI/ML into NII systems.