Physical Optics Corporation — Department of Defense SBIR Phase I: N181-029

Description

To address the Navys need for a maritime Automatic Target Recognition (ATR) System using ISAR data for the Tomahawk Missile, Physical Optics Corporation (POC) proposes to develop a new ATR System, Automated Identification of Maritime ISAR Targets (AIMIT). It is based on modern deep-learning techniques for identification of objects and salient features in image data. Specifically, the innovation in using deep- learning image recognition algorithms for maritime ISAR targets will enable the Tomahawk Missile navigation component to accurately target hostile ships, while avoiding innocent bystanders. Additionally, AIMIT deep-learning algorithms are computationally efficient at runtime and updatable via parameter files. As a result, this technology offers a high probability of correct identification, computational throughput sufficient for a 5 Hz update rate, and the capability to readily update the system with new targets. It directly addresses the Tomahawk Missile Program requirements for SWaP, update rate, and classification accuracy. In Phase I, POC will demonstrate the feasibility of AIMIT by demonstrating the performance against either a synthetic or experimental dataset, reaching TRL-3. In Phase II, POC plans to develop a prototype to show that the performance of AIMIT against representative data is acceptably high, bringing the technology to TRL-4.