IN Brief:
- RTX’s BBN Technologies has a DARPA XENA award for long-standoff X-ray analysis.
- The programme targets transmission X-ray scenarios out to near-kilometre ranges.
- Algorithm development will have to translate into rugged, fieldable compute and test.
RTX’s BBN Technologies has been awarded a contract under DARPA’s X-ray Extreme-range Non-imaging Analysis (XENA) programme to develop long-range X-ray imaging algorithms intended to infer the hidden geometry of man-made objects from distances approaching a kilometre.
The work is aimed at situations where close access is unsafe, impractical, or denied, with potential applications spanning threat assessment and emergency response. “Long-range X-ray imaging requires a fundamentally different approach,” said Joshua Fasching, principal investigator at BBN Technologies.
DARPA’s XENA programme is framed as an algorithmic push to make useful inferences from long-standoff transmission X-ray data when motion blur is present and there is no prior information about what is inside the object being imaged. The agency describes a step-change goal: extending the state of the art in transmission X-ray from single-metre ranges to single-kilometre ranges, while working at hard X-ray energies at or above 150 keV.
BBN said its approach will combine advanced mathematical modelling and image analysis to enhance visibility even when the data is incomplete or noisy, and without relying on large training datasets. The team will run simulations, build software, and test performance against scenarios designed to replicate the weak signals, limited viewpoints, and blur that can overwhelm conventional imaging methods at long stand-off distances.
BBN is leading the work with the Georgia Institute of Technology, with activity taking place in Cambridge, Massachusetts, and Atlanta, Georgia.
Long-range transmission X-ray is an awkward regime for traditional techniques because the imaging chain is fundamentally photon-limited. At stand-off distances, attenuation through air, geometry constraints, and short exposure windows can yield sparse, low-contrast measurements, while even modest platform motion can smear the signal enough to bury edges and internal features.
XENA’s emphasis on blind processing, without prior knowledge of interior composition, pushes performers towards methods that can exploit shared structure across a small set of views, rather than relying on dense scan angles or carefully controlled calibration common in industrial and medical computed tomography. For BBN, the target is operational inference: extracting actionable geometry when the input is closer to a handful of compromised snapshots than a clean scan.
If the algorithms are to transition beyond a laboratory demonstration, the industrial work starts by packaging compute and verification into something repeatable. Field deployment typically demands deterministic processing, hardened electronics, and stable performance across changing sensors and platforms, which is an engineering and production challenge as much as a modelling one.
Hard X-ray scenarios also carry practical constraints that shape any downstream manufacturing plan. Shielding and safety interlocks influence size, weight, and integration, while detector availability, calibration routines, and environmental robustness drive supply chain choices and acceptance testing. Even for an algorithm-led programme, those realities sit in the background, because they decide whether the eventual system can be reliably built and validated at scale rather than treated as a one-off capability.
