Developing the behaviours for use in RoboCup 2015 based on rUNSWift architecture.

Abstract

The RoboCup Standard Platform League has two teams, each consisting of five robots play football against each other in a semi-controlled setting. The robots used have the same hardware and modifications are not allowed. The purpose of this thesis was to find a method to improve the overall performance displayed during 2014 RoboCup and implement the method(s). During the course of the project, a new codebase, developed by team rUNSWift, was evaluated, tested and then adopted as it offered improvements compared to the Austin Villa codebase used in 2014. As the codebase offered only basic core functionality, a behaviour module needed to be implemented to offer both low- and high-level behaviours. The behaviours developed provide low-level functionality for movement, ball alignment and targeting and high-level functionality for basic soccer gameplay according to RoboCup 2015 rules. The individual strategy mimics the system used in 2014 with the main difference being the ability to recognize our teammates and then use that information to avoid collisions while trying to hit a ball that is in the common playing area of the two robots. The kick and walk performance appear more stable, as they are both dynamically generated using rUNSWift’s motion system. The walk is also offers greater configurability and needs careful calibration for tuning the input parameters.