Data Fusion Handoff

Period of Performance: 11/08/2006 - 05/08/2007

$79.8K

Phase 1 SBIR

Recipient Firm

Adaptive Methods, Inc.
5860 Trinity Parkway Array
Centreville, VA 20120
Principal Investigator

Abstract

Multi-sensor data fusion is a critical technology used to support contact tracking and association for many Navy missions and systems. Traditional approaches focus on fusing contact position and kinematics data from multiple heterogeneous sensors, with overlapping coverage areas, in order to provide more reliable estimates of the contact s state. However, these systems fail to meet the requirements for emerging Navy missions, such as the National Maritime Domain Awareness (MDA) goal for persistent surveillance of surface vessels. For MDA and other programs it is critical to be able to reliably associate contacts from one sensor coverage region to another, even with multiple and possibly crossing tracks. In this proposal, an approach for multi-sensor multi-target fusion with persistent track identification numbers is presented. Contact association using Sequential Bayesian Iteration and Inference is discussed and proposed for development in the Phase I work. Development of a particle filter approach for fusion of non-overlapping sensor data is also proposed. Integration of these fusion approaches with a method for propagating track numbers from the sensor, platform, group, theater, and world view-point is also presented. A proof of concept non-real-time simulation of the approach is also proposed.