Adaptive Battle Agents: Complex Adaptive Combat Models

摘要

This work explores emergent behaviour in a complex adaptive system, specifically an agent-based battlefield simulation model. We explore the changes in agent attribute sets through the use of genetic algorithms over a series of battles, with performance measured by a number of different statistics including number of casualties, number of enemy agents killed, and success rate at "capturing the flag". The agents' capabilities include (but are not limited to) manoeuvrability upon the battlefield, formulating, sending, receiving and acting upon messages and attacking enemy agents.