Novel method of estimating moving target spatial dynamics for radar imaging

The increasing proliferation of mobile missile technology, as well as the adoption of CONOPs that hide launchers between deployments, are driving a need for the capability to identify fleeting targets while they are in motion.   Radar sensors are essential for this capability, as other sensors cannot perform in all weather and light conditions.  Because vehicles move straight the vast majority of the time, but with inadvertent irregular translational and rotational motions defocus the vehicle’s radar image, estimation of moving vehicle spatial dynamics is entwined with vehicle identification. EMSI proposes to extract moving vehicle spatial dynamics by combining traditional radar processing with an adaptation of state-of-the-art computer vision structure-from-motion processing to multi-radar data, and to define radar topologies and waveform parameters suitable for extracting the dynamics.  All training in Phase I and the bulk of the training thereafter will rely on synthetic data.   The speed of EMSI’s STARDUST synthetic generation service enables the generation of sufficient training data. .