AI Lecture Series Talk - Michael Buro - Planning in RTS Games

Time: 12:00-13:00, Wednesday March 4th,

Place: K5, Kringlan 1.

ABSTRACT:

In RTS games, players struggle over resources that are scattered on an initially uncharted terrain in real-time. Having access to more resources allows them to build more forces quicker and to finally overcome opponents by destroying their units and structures. There are many challenges in creating AI systems that can reach or outperform human players in this popular game genre, ranging from multi-unit pathfinding, over squad combat tactics and scouting, to plan-recognition, opponent modeling, and army deployment. RTS games are therefore ideal applications for research into real-time AI.  In this presentation I will give an overview of my group's past and ongoing work on this exciting research problem. I will also describe our RTS game engine ORTS which is being used in the annual RTS game AI competition. My hope with this talk is to spark interest in RTS game planning problems and the ORTS competition among planning researchers.