A Comparison of Computational Cognitive Models: Agent-Based Systems Versus Rule-Based Architectures

Abstract

Increased operational costs and reductions in force size are two of the major factors driving the need for improved computer simulations within the military community. Human performance models are used in various aspects of simulation, including the control of computer-generated forces, as tactical decision aides, in intelligent tutoring systems, and in new system design. This research compares two categories of human performance models: multi-agent systems and rule-based architectures. Each model has its own strengths and weaknesses and is best suited for certain applications. Complex military simulations need human performance models that take advantage of the strengths of more than one type of model. This study compares the implementation and performance of these two models, and demonstrates the need for hybrid systems that employ the best aspects of each for a given situation. The thesis is organized as follows: Chapter II discusses the background of naturalistic decision making and provides overviews of the following computational cognitive architectures: Adaptive Control of Thought-Rational (ACT-R); Soar; Cognition as a Network of Tasks (COGNET); Java Expert System Shell (JESS); Operator Model Arhitecture (OMAR); Micro Saint; Neural Networks; and Agent-Based Systems. In Chapter III, the two models are applied to the decision making problems of the Officer of the Deck onboard a submarine as he maneuvers the vessel to carry out the assigned mission. His highest goals are to avoid running aground, to avoid colliding with another vessel or stationary navigation hazard, to remain within the assigned operating area boundaries, and when these safety goals are met, to carry out the assigned mission. Chapter IV discusses the application of the agent-based model to the problem, and chapter V discusses the application of the rule-based model to the problem. Chapter VI compares the two models. Chapter VII provides conclusions and recommendations for future work. (9 re7

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 2003
Accession Number
ADA415186

Entities

People

  • Craig L. Oeltjen

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • C4I
  • Ground and Sea Platforms
  • Weapons Technologies

DTIC Thesaurus Topics

  • Cognition
  • Combat Simulations
  • Computational Science
  • Computer Languages
  • Computer Programming
  • Computers
  • Coordinate Systems
  • Expert Systems
  • Human Behavior
  • Hybrid Systems
  • Information Systems
  • Military Research
  • Motor Skills
  • Multiagent Systems
  • Seabed
  • Self Organizing Systems
  • Simulations

Readers

  • Agent-Based Social Robotics and Mobile-Assisted Learning in Virtual Environments.
  • Business Analytics
  • Team-Based Human-Centered Cognitive Task Decision Making and Information Performance.

Technology Areas

  • AI & ML
  • Fully Networked C3
  • Fully Networked C3 - Command and Control