Sub-Wavelength All Motions Information (SWAMI)

In the realm of SAR imaging, a critical challenge involves the identification and motion characterization of non-cooperative moving vehicles. This challenge, at a fundamental level, requires the extrapolation of 3D information from 1D radar signals. This is only possible provided the application of principled assumptions regarding the structure of the incoming data and limitations of physics that constrain the manner of target motion. The historical precedence of methods relying on tracking reflections from scattering centers that persist throughout the signal history data for extraction of target geometry and motion has yielded some success. But the examples of success have not yet been sufficient to motivate frequent application in operational systems. This has only begged the question of whether or not novel methods that consider the entire signal content may yield downstream improvements. To that end, MTRI, Inc. and MTU will conduct the Phase II SBIR Sub-Wavelength All Motions Information (SWAMI). This project will investigate the feasibility of improving radar algorithm development to generate accurate motion estimates for 2D and 3D radar imaging, enhance ground clutter suppression methods, and increase the robustness of the radar system to the environment and radar system parameters. Phase I of SWAMI has demonstrated that significant gains can be made in the realm of target motion estimation and 3D imaging. As part of Phase II, we will further develop the imaging capabilities by analyzing the performance of the algorithms in relation to SAR collection duration, target motion dynamics, and vehicle environment in order to develop a quantification of algorithm performance uncertainty. By doing so, the proficiency of the algorithms will be demonstrated under a variety of conditions, showing that it is reliable for practical SAR data collections. These improvements result from the application of new insights from differential geometry, optical flow experiments and visualization methods, and extensions of our previous work using invariant theory.