Non-communicative multi-robot coordination in dynamic environments

摘要: Within a group of cooperating agents the decision making an individual agent depends on actions other agents. In dynamic environments, these dependencies will change rapidly as result continuously changing state. Via context-speciflc decomposition problem into smaller subproblems, coordination graphs ofier scalable solutions to multiagent making. this work, we apply continuous (robotic) domain by assigning roles and then coordinating difierent roles. Moreover, demonstrate that, with some additional assumptions, can predict agents, rendering communication super∞uous. We have successfully implemented proposed method our UvA Trilearn simulated robot soccer team which won RoboCup-2003 World Championship in Padova, Italy.