Strategy selection, goal generation, and role assignment in a robotic soccer team

Abstract

This paper describes a method of strategy selection, goal generation and role assignment for robotic soccer teams. This method has been implemented in the All Boltz, a RoboCup team from the University of Auckland. Firstly, an offensive and defensive value for a position is computed using an offensive and defensive objective function. The system then selects either to play an offensive or defensive strategy. Based on the selected strategy, some potential goals are generated for the robots. Goals consist of position and orientation information. The system selects five goals which maximize the strategic objective function. Each robot then selects a goal from the set of generated goals and develops a plan to achieve this goal. Note that all computations in the described method are local and can thus be applied to a distributed system, where each robot makes its own decisions.