| Being the main research background of MAS (Multi-Agent System) and DAI (Distributed Artificial Intelligence), RoboCup (The Robot Soccer World Cup) is an international organization of research and education. RoboCup supplies a standard tournament bed to foster and test the research and development of Robotics and Artificial Intelligence.Putting Successful Defence in RoboCup Simulation as research target of MAS, this dissertation sets up a set of defence strategy, in that puts forward three defence strategies respectively on the idea of whole defence, part defence and individual defence: the Defence model based on switching formation, the Defence model of Backfielder based on SBD strategy and the Goalie Defence model based on alterable goal line.The Defence model based on switching formation applies Case-based learning to design formations breaching the experiential parameter method. And this model focuses on the function in defence of formation, and enhances the defensive capability of the whole team observably.Studying collaboration of MAS, the paper brings forward SBD (Strategy Based on Disorganization) for disorganization. The Defence model of Backfielder based on SBD strategy as active defence action can be applied easily after the backfielder intercepting the ball. The model can disorganize the collaborative actions of opponents efficiently, and then lighten the pressure of defensing the goal by a long way.By researching and analyzing Defending the Goal Line, the paper puts forward alterable goal line, and also researches two methods: the alterable parallel goal line and the alterable across goal line, that can display the defensive effect of the goalie sufficiently.Adopting the defence strategies advanced in this article, CSU_YunLu2003, which participated in the Chinese Robot Competition '2003, has been awarded second prize. The results of actual tournament show that the defence strategies are feasible and highly effective. |
PDF Full Text Request