OUR MISSION IS TO SUPPORT THE WARFIGHTER,
THROUGH SCIENCE
Technology supports innovation, but it is only when dedicated individuals are brought together who all thrive on a culture of developing solutions to difficult real-world challenges that real innovation occurs. Members of AER’s newly established Target Recognition and Advanced Exploitation (TRAX) Division bring together a range of talents and skills with the common thread that they embrace challenges and find satisfaction in working together as a team to develop new and innovative ways to address them.
While new to AER, the TRAX Division is comprised of scientists and engineers with a collective background and experience in target recognition and related fields of study that cover many decades. This includes a comprehensive understanding of data collection methods and phenomenology across multiple sensor modalities. Target recognition solutions designed and developed by TRAX employ a combination of photogrammetry, computer-vision, and machine learning approaches that are backed by a solid understanding of physics-based target and materials attributes.
TRAX’s capabilities span the electromagnetic spectrum, processing sensor data to detect and identify targets beginning in the visible region and continuing through the infrared and into the microwave regions. Each sensor modality provides its own benefits relating to its ability to detect and identify targets. The diversity of TRAX’s capabilities enables its members to leverage the benefits of each sensor modality when designing Multi-INT solutions to meet or exceed operational requirements.
Synthetic Aperture Radar (SAR) provides several benefits over electro-optical (EO) and infrared (IR) sensors as they mitigate several limitations of those sensors. The frequencies in which SAR sensors operate enable their signal to pass through clouds rather than to be reflected by them, as is the case with EO and IR sensors. This attribute enables SAR to be employed as an all-weather capability. Since they are active sensors, providing their own energy to illuminate targets, they may also be employed to support night operations when solar illumination is not present.
The most obvious limitation of images formed from SAR data is that they do not have the appearance of traditional images collected by optical sensors. As a result, the expertise of a trained analyst is typically required to extract the valuable information content that they contain. An alternative means to extract this information is to employ automated target recognition (ATR) approaches. These approaches are automated in the sense that they detect and identify targets in collected image data by means of algorithms implemented in software rather than manually by analysts. As a result, ATR approaches make it possible for a broader range of end users making time-critical decisions to have access to this valuable information.
RAPTOR (Rapid Target Predictor) is a TRAX-developed technology that is central to TRAX’s family of capabilities. RAPTOR is employed to rapidly generate target predictions corresponding to their representation in EO, IR, and SAR sensor modalities. These predictions are used as input to TRAX’s model-based ATR approaches and to rapidly generate vast amounts of labeled, high quality simulated target signature data to train TRAX’s machine learning ATR approaches. RAPTOR employs 3D CAD models and sensor parameters, coupled with illumination and thermal transfer models, to generate the predicted target signatures.
TARGET PREDICTIONS AND SIMULATED TARGET CHIPS GENERATED BY RAPTOR
Model-based approaches provide end users with the ability to make a visual comparison of predicted target signatures to detected targets in sensor data. Rather than simply accepting the target label provided by ATR approaches, a visual comparison provides the user with a means to either accept the target label with confidence or to dismiss the call. In the case of SAR data, the capability also provides a means to either justify or dismiss a target call that may not have been otherwise obvious. This capability is particularly beneficial when the end user has not been formally trained in SAR data phenomenology.
SAR TARGET PREDICTION MATCHED TO COLLECTED SAR IMAGE SCENE
TARGET SEARCH FOR A-4 AND A-7 AIRCRAFT LOCATED ON USS MIDWAY IN SAN DIEGO, CA
The TRAX team and its family of capabilities support both defense and intelligence needs, including those of the Intelligence, Surveillance and Reconnaissance (ISR) and Combat ID communities. Its capabilities are applicable to a broad range of airborne and space-based sensor platforms and provide cost-effective, high performance, sustainable solutions to those with time-critical target recognition mandates. The newly established TRAX Division continues the tradition of scientific excellence at AER, with a team that is dedicated to the development of target recognition capabilities that are designed to address both the needs and challenges of today as well as those of tomorrow.