Multi-Agent Cooperation Based on Opponent's Behavior Forecast

It is necessary that forecasting the opponent's behavior effectively and exactly and then adopting corresponding strategies for MAS under dynamic environments to win in a competition. This paper brings forwards a new method of opponent's behavior forecast and realizes the multi-agent cooperation using this method. Finally, it is proved that the method is feasible and effective in RoboCup simulate competition system.