Abstract:

The article is devoted to the modelling and interactive simulation of combat units activities for staff games. The formation of units (computer generated forces) is considered as multi-agent network, where any agent who demonstrates intelligent behaviour on the battlefield. The intelligence of the agent is considered in the aspect of terrain overcoming and decision making process. The paper represents the idea of extending the combat simulator by input of machine learning mechanisms. The method of knowledge acquisition (course of actions generation by commander in the interactive way) is presented for a number of cases. Selected cases are presented here for reinforcement learning purposes. The development of a method for learning and then realising conflict scenarios with using of an intelligent agent (algorithm) against a human or an agent against an agent will be considered.