Fusion of Kinematic and Identification (ID) Information

ABSTRACT: To improve the tracking and pattern analysis of high value targets (HVTs), a technique for associating and fusing large quantities of kinematic data with temporally sparse, but highly informative, target ID information is required. Specifically, there is a need to develop a data association algorithm that will connect highly informative target ID observations with kinematic tracks in order to estimate the set of likely trajectories that an HVT could have traveled between these ID observations. The primary challenges include: (1) ambiguity and gaps in tracks over extended time periods, (2) high rate and volume of kinematic data to be processed, and (3) use of multi-INT classification data with varying feature spaces, uncertainty models, and ambiguities. Lakota Technical Solutions, Inc. proposes the Boundary Value and Retrospective Correlator (BVARC) algorithm for association of multi-INT classification and kinematic data with temporally sparse ID information. This approach provides an O(N) storage size scalability for track hypotheses, O(N log(N)) computational complexity for data association, and leverages a Technology Readiness Level (TRL) 6 technology to quantify the consistency and conflict between multi-INT classification data. The BVARC provides a tractable capability to identify and rank the set of trajectories traveled by an HVT for large-scale tracking problems.; BENEFIT: The DoD encounters many problems that require the tracking, classification, and analysis of target behavior over extended time periods far beyond the update intervals of a sensor, including: (a) high-value targets traveling in vehicles on land, (b) ballistic missile handoff between over-the-horizon (OTH) sensors, (c) maritime vessels traveling great distances with large gaps in sensor coverage, (d) submarines with intermittent detections, and (e) orbiting satellites that performing maneuvers during large gaps in sensor coverage. Each of these examples requires the ability to maintain and evaluate trajectory hypotheses over extended periods of time with feasible storage and computational requirements. Lakotas proposed BVARC is a generalized approach that can be employed in all of these domains with computational feasibility in large-scale tracking environments where large numbers of hypotheses are inevitable. As a result, BVARC has a large canvas of potential transition targets within the Air Force, other DoD, and potentially private commercial domains.