Cognitive Fusion

Period of Performance: 06/18/2008 - 06/18/2010

$1.69MM

Phase 2 SBIR

Recipient Firm

Soar Technology, Inc.
3600 Green Court Array
Ann Arbor, MI 48105
Principal Investigator

Abstract

Providing a dominant response to terrorism and insurgent-driven asymmetric wars is a complex problem and is accelerating demand for high-quality multi-INT fused Intelligence, Surveillance, and Reconnaissance (ISR) products. Soar Technology, Inc. and the Lockheed Martin Co. propose to significantly advance the state of the art in level 1 multi-INT data fusion by developing a novel technology that will significantly extend the length of high purity, multi-INT, fusion tracks for both real-time and forensic applications. The proposed Cognitive Fusion technology is a hybrid fusion system that provides long tracks at high purity with high confidence. We will achieve this innovation by integrating quantitative observations (such as those produced by typical kinematic trackers) and qualitative abstractions of the environment. This hybrid computation is achieved by balancing relatively slow, but contextually rich, cognitive agent reasoning with a fast, but knowledge poor, fusion algorithm. An agent imbued with rich knowledge of the domain, mission, and analyst hypotheses will bring more knowledge and algorithms to multi-INT fusion in comparison to strictly quantitative kinematic processors. Our primary technical deliverable will be a hybrid fusion system for real-time and forensic multi-track correlation and data association. The centerpiece of this fusion system will be a Cognitive Fusion Module that incorporates a software agent and data manager to implement a novel knowledge representation and reasoning algorithm for adversarial behavior recognition.