Developing Realistic Cooperative Behaviors for Autonomous Agents in Air Combat Simulation
Abstract
This thesis investigated, developed and implemented cooperative decision-making behaviors in an air combat simulation by using a knowledge-based system. Knowledge-based systems were well suited for this task because of built- in features such as inference engines and rule-based constructs. This thesis addresses the specific problem of generating autonomous forces for inclusion in the Advanced Research Projects Agency Distributed Interactive Simulation program. Existing autonomous forces implementations lacked flexibility, realistic behaviors, real-time planning and other features. The simulation system in this thesis addresses the problem of realistic behavior by modeling pilot decision processes rather than aircraft platforms. The system is based on phased control of a blackboard architecture. Modular knowledge bases partition rules to process decision data. Cooperative behaviors are based on a leader- follower relationship. Agents share the workload in assessing threats. Leaders make the initial decision, but followers react independently if necessary. The simulator described in this thesis provides an architecture and design for modeling combat pilot decision processes. The system was developed using the C Language Integrated Production System Object Oriented Language.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1993
- Accession Number
- ADA274077
Entities
People
- Dean P. Hipwell
Organizations
- Air Force Institute of Technology