Accession Number:



Advanced Agent Methods in Adversarial Environment

Descriptive Note:

Final rept. 14 Jun 2004-14 Dec 2005

Corporate Author:


Report Date:


Pagination or Media Count:



The suggested research effort aims at continuation of the research in the field of social knowledge, social behavior and meta-reasoning in multi-agent systems. The main and fundamental theme of this research project is extending meta-reasoning and social knowledge manipulation to the problem of intent modeling in the non-trivial domains of non-cooperative, competitive, i.e. adversarial behavior. The main ambition is to extend meta-reasoning for the problem of intent modeling in the non-cooperative, cooperative, i.e. primarily adversarial domain. This effort retains the focus on environments with partial communication inaccessibility. This extension of the metareasoning to adversarial agent behavior is a natural step in applicability of this technology to Information Operations. Dr. Pechoucek has a long and successful history with AFRL working under grants to Czech Tech University with Dr. Vladimir Marik. Deliverable 1 month 6 Interim report defining the concept of non-collaboratively and hostility in multi-agent systems. with design of a scenario for investigating agents social behaviour in non-collaborative and adversarial environment. Deliverable 2 month 12 Interim report describing potential role of agents acquaintance model, social knowledge and meta-reasoning in non-collaborative environment with partial communication accessibility. Deliverable 3 month 18 Comprehensive final report providing technical description of the entire research effort including formal definition of measures and quantities of agents collective behaviour, adapted social knowledge model, meta-reasoning model and coalition formation algorithms for non-collaborative and adversarial environments and proof-of-concept experimental verification.

Subject Categories:

  • Cybernetics
  • Psychology
  • Command, Control and Communications Systems

Distribution Statement: