ONR C-IED STIFLE (Stigmergic Tracking of IED Factories, Locations and Events)

摘要

The STIFLE project, funded by the ONR Counter-IED Basic Research program, has three major objectives, which are reflected in the three main tracks of our project execution: (1) 'Enhanced Representations' Track: Extend the predictive polyagent modeling construct to include explicit reasoning over task execution by individuals and groups; (2) 'Model Analysis' Track: Develop theoretical, formal and experimental analysis tools and methods to characterize and influence the dynamics of predictive polyagent models; (3) 'IED Prediction Prototype' Track: Apply the extended modeling and analysis capabilities to the problem of IED prediction and forensics Towards the first objective, collaborated with Prof. Keith Decker (University of Delaware) to integrate our polyagent modeling approach within the TAEMS framework, a formal representation and reasoning mechanism for hierarchical task networks. Together with Prof. Bob Savit (University of Michigan), we explored various approaches to formally describe and analyze our predictive polyagent models in support of the second objective. Finally, our development and experimental analysis of alternative polyagent models of IED emplacement (based on initial models and a framework supported by the DARPA RAID adversarial reasoning module) supported the third objective.