Leveraging parallel architectures in AI search algorithms for games

Abstract

This document contains a summary of research performed by the author on the topic of search algorithms for games. An outline of the problems being addressed is provided, along with the progress already made, and planned future work. The specific subjects studied are: parallelizing UCT search on GPUs, the development of a hierarchical search framework for Real-Time Strategy (RTS) games and the building placement problem in RTS games. We propose to take advantage of different parallel architectures to help solve these problems.