Human Behavior Representation - Recent Developments
Abstract
Until recently, the command and control processes and especially the decision process were - if at all - hard-coded in the closed combat simulation models. Although in other application areas expert systems and rule systems as well as neural nets were used to find solutions to similar problems, in the closed combat community these approaches are still not used sufficiently. The problem with hard-coded decisions is always, that they are evaluated only for a special environment. Being transferred to another environment the decision model is seldom capable to meet the new requirements. Self-optimization and the skill to adapt to new situations are nowadays more often required by such systems. This lecture introduces adjustable rule sets and neural networks as alternatives to generate orders in a given situation. The results and the possibilities for further improvement and evaluation of the results are given. In addition, some technical aspects of coupling the results in form of CGF federates with legacy systems and the challenge of structural variances are copied. Further actual developments in the modeling of Human Behavior Representation come from Germany, where a new approach trying to catch the insight from latest psychology findings as well as modern agent technology are merged to a new solution of the well-known problems.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2003
- Accession Number
- ADA485081
Entities
People
- Andreas Tolk
Organizations
- Old Dominion University