A cooperative CiberMouse@RTSS08 team

Abstract

This paper describes the strategy and implementation details of a cooperative agent team for the CiberMouse@RTSS08, a robotics simulation competition. The paper introduces several concepts concerning cooperative robotics, giving an overview of its applications and challenges. It also presents the CiberMouse@RTSS08 competition rules and the associated simulation system. The proposed approach is based on a deliberative architecture, thus providing an autonomous intelligent behaviour. A probabilistic mapping procedure, based on the Bayes' theorem, is used to maintain an internal world state. Quadtrees and A*are used for path planning and plan execution. A specific methodology was developed for beacon finding. The agents cooperate on exchanging world and beacon information. In order to overcome the simulator's broadcasting distance limitation, a communication network between the agents was created. The paper analyses the impact of cooperation on this particular problem, by comparing the team's performance in four different situations: communication with network and beacon exchanging, communication without network, communication without beacon exchanging and no communication at all. © 2009 Springer Berlin Heidelberg.