Using automatic speech recognition models, SwRI computer scientists are developing methods to transcribe collected communications signals for faster, easier analysis. This technique reduces operator workload, increases audio transcription, improves accuracy and enables new features like keyword matches. Researchers are fine-tuning the model using distorted signal data, including fading noise, impulse noise and varying signal-to-noise ratios. The model has been integrated into SwRI’s Frontier system, a signal evaluation tool. SwRI is considering expanding the speech recognition tool by adding language and speaker identification to improve model performance.

SwRI developed and deployed a high-performance, high-frequency direction-finding (DF) system to allow users to discover new or elusive signals of interest using arrival direction as a filtering criterion. The radio frequency spectrum between 3 and 30 megahertz is congested with signals that can propagate for thousands of miles. SwRI’s newly deployed process, Full Spectrum DF, significantly accelerates DF processing to more than 33 million calculations per second. The large volume of calculations allows angle-of-arrival measurements on very short signals and signals in a crowded spectrum, providing more precise information about the location of a signal’s source.

Turning forensic engineering into mission assurance, SwRI uncovered the root cause of multiple B-52 alternating current generator failures that caused catastrophic outcomes in the field, including near-crashes. Engineers captured 3D scans at micrometer levels to assess the geometry of key generator parts, bearing components and rotor inertial properties. Subtle wear patterns and misalignments emerged under digital scrutiny. Using scanning electron microscopy imaging and X-ray diffraction, SwRI investigators discovered microcracks, pitting and degraded lubrication pathways — previously undetected signs of fatigue and contamination.

For decades, SwRI has worked with the Air Force to extend the life of aging military aircraft. This year, we received new contracts to sustain the A-10 Thunderbolt II attack aircraft, the T-38 Talon supersonic trainer, the C-5 cargo carrier and the B-52 Stratofortress bomber, among others. These fleets were introduced between the 1950s and 1970s. SwRI will also provide technical engineering support for smaller fleets managed by Hill Air Force Base, including the T-41 and T-52 trainers and the E-9 surveillance aircraft.

SwRI invested in the Air Force community of Warner Robins, Georgia, to advance national defense technology, with the first SwRI-owned property outside of Texas. After decades of leasing there, SwRI opened a 33,000-square-foot, $18.5 million building strategically located 3 miles from Robins Air Force Base. The new facility features conference rooms, offices, labs and equipment to support development of EW technologies that detect and intercept enemy radar signals. Engineers use the facility’s server equipment to train highly complex AI algorithms locally, eliminating the lengthy process of copying data sets between our San Antonio headquarters and Warner Robins.

Image

SwRI developed algorithms for event cameras, digital imagers that record pixel changes within a field of view, to track small, fast and agile uncrewed aerial systems (UASs). These counter-UAS technologies support airspace security.

In 2025, SwRI developed software to help the U.S. Department of Defense identify and respond to chemical and biological threats. The tool helps make agile medical decisions about countermeasures and other actions to rapidly address a danger or threat.

In 2025, SwRI expanded the scope of the Laser-Induced Particle Impact Test (LIPIT), dramatically increasing the precision and efficiency of material ballistic resistance testing at scales not possible before. Previously, LIPIT only supported testing using extremely small projectiles. To make LIPIT results more applicable to full-scale ballistics, SwRI engineered a test system to launch larger projectiles of 0.3 millimeters in size. Further, the system was automated to accelerate from a manual test rate of 30 to 40 tests per day to the automated rate of 200 tests per hour.

In 2025, cybersecurity work included penetration testing, risk assessment and secure software development for critical systems. In 2025, SwRI’s team identified cybersecurity vulnerabilities in both 240-volt Level 2 electric vehicle (EV) chargers and direct current fast chargers (DCFC). As a result of the Level 2 work, the EV charger manufacturer published a public advisory for 76 impacted products, and the Cybersecurity and Infrastructure Security Agency issued two common vulnerability and exposure (CVE) reports. The DCFC research uncovered a critical cybersecurity risk with global implications, leading to another CVE report.