Net-centric ACT-R-Based Cognitive Architecture with DEVS Unified Process

摘要

Air Force Research Lab (AFRL) research efforts employing cognitive and behavioral modeling are growing in scope and complexity as they work to integrate models into larger distributed systems as cognitive agents, synthetic teammates or human operator surrogates. Efforts to transition cognitive modeling from the laboratory to operational environments are stymied by the isolated nature of current cognitive modeling software tools that are not readily extensible, interoperable or scalable e.g. ACT-R. In this paper, we describe an attempt to build a component-based architecture using the Discrete Event Systems Unified Process (DUNIP) on a distributed net-centric platform that eliminates these impediments. We show how the ACT-R architecture is extensible and can serve as a component in larger net-centric systems of systems frameworks such as Department of Defense Architecture Framework. We will also address the issue of platform independent modeling and how Domain Specific Languages (DSLs) can be integrated within DUNIP. We then demonstrate how developing the architecture and related software infrastructure in DUNIP gives it net-centric capabilities that support large-scale integration.