OxBlue2008(2D) Team Description
Jie Ma · 2009
Abstract. OxBlue2009 (2D) is a robot football team for RoboCup 2D simulation. In this paper, the decision structure of our team will be presented. Under this structure the fundamental cooperative behaviour formation is briefly presented. It is accompanied by a instance-based learning method that approximates opponent formation functions. As to the high-level cooperation, PSP (Policy Search Planning) that we proposed in 2008 is reviewed. Policy Search is used to find an optimal policy for selecting plans from a plan pool; it extends an existing gradient-search method (GPOMDP) to a MAS domain. 1