Multi-Robot Cooperation in RoboCup-SPL

Abstract

RoboCup Standard Platform League (SPL) is an international robotics soccer competition with an aim to foster research in artificial intelligence and its application in advance robotics. It involves multi-robot cooperation to achieve a common team objective. Hence the need of dynamic task allocation in real time, under a partially known and dynamic environment, arises to achieve a global objective through cooperation or task decomposition. Multi-robot cooperation has gained some attention in the recent past, but to-date only a handful of architectures exist. However, most of the research is ad-hoc and focused on developing proof of concept of available architecture using simulation based techniques, while rarely being employed on actual hardware platforms. This research proposes a robust hybrid architecture for multi-robot cooperation that exploits the advantages of both market based and behavior based architectures, which was employed for RoboCup-SPL 2016 by Team-NUST.