Advanced Iterative Action Planning for Intelligent Real-Time Agents

摘要

Action planning by intelligent agents in open dynamic multi-agent worlds is a challenging problem when creating autonomous robots, UAVs and other autonomous systems. A method for planning actions in an ever-changing multi-agent world is proposed, that ensures the effective use of the deliberation time by planning mental actions. Knowledge representation models necessary for the agent to implement advanced iterative planning are considered. The proposed approach is illustrated by examples of planning an agent’s actions in virtual soccer.