Related Experiment Videos
FOX-GA: a genetic algorithm for generating and analyzing battlefield courses of action.
J L Schlabach1, C C Hayes, D E Goldberg
1Technology Integration Office, Office of Deputy Chief of Staff for Intelligence, Department of the Army, Pentagon, VA, USA. jlschla@vulcan.belvoir.army.mil
Evolutionary Computation
|April 13, 1999
Summary
This study introduces FOX-GA, a genetic algorithm (GA) for military maneuver planning. It overcomes slow simulation times with an efficient evaluator and ensures diverse plan options using niching strategies for better decision support.
Area of Science:
- Artificial Intelligence
- Computational Intelligence
- Military Science
Background:
- Military maneuver planning presents complex decision-support challenges.
- Traditional plan evaluation methods are computationally prohibitive for genetic algorithms (GAs).
- Existing GAs often lack solution diversity, limiting user choice.
Purpose of the Study:
- To demonstrate the efficacy of genetic algorithm (GA) technology in complex military maneuver planning.
- To identify key properties of GA solutions for effective decision support in complex domains.
- To address the computational and diversity limitations of GAs in planning.
Main Methods:
- Developed FOX-GA, a genetic algorithm tailored for military maneuver planning.
- Implemented an efficient, course-grained "wargamer" evaluator to balance speed and accuracy.
- Integrated a niching strategy into the GA's selection mechanism to promote solution diversity.
Main Results:
- FOX-GA effectively generates and evaluates maneuver plans within practical time constraints.
- The efficient evaluator significantly reduces plan assessment time compared to detailed simulations.
- The niching strategy successfully produced a diverse set of distinct plan options.
Conclusions:
- FOX-GA offers a viable and efficient GA-based solution for military maneuver planning.
- The approach enhances decision support for time-constrained military personnel.
- This work provides insights into applying GAs to complex, real-world planning problems.