The Research Of Multi-artificial Life Body's Behavior Collaboration Based On Robot-soccer

Soccer robot is a value application example in the artificial life, in which the soccer robot is a typical example of the robot, in this paper, the soccer robot is as the research platform of multi-artificial life body. The soccer robot system is a mobile robot system and its collaboration is very important, therefore, the robot soccer game is a typical coordination multi-robot system searching on artificial life. In a number of soccer robot system, each robot soccer can be seen as an artificial life form, the behavior collaboration of soccer robot can be achieved through the decision-making subsystem of the four systems in the robot soccer.In the decision-making subsystem, this paper will use the three-tier model of high layer-strategic level,middle layer-role distribution layer and low layer-the robot action execution layer. At the strategic level, the prediction method is used to improve the behavior collaboration, fuzzy Petri method is used to research the high-level behavior collaboration. In the low level of the decision-making subsystem, the improved potential field is used to research the low-level behavior collaboration. In the support of research project by relying, this article views the simulated robot soccer as a research platform, expands this research work of the collaborative behavior of artificial life. This paper is divided into the following three points:First, the AR model is established according to the order determination criterion AIC and the correlation function of the sample then the position of the ball is predicted by using it, the result of prediction error is small. This method predicts the next situation based on its previous cycle situation and improves the ability of the next behavior collaboration,Second, the situation of the robot soccer competition area is instantaneous variable, the strategies that can be choosed in each cycle are variety, the main research of improving the collaboration level of multiple soccer robots is to think how to choose the one of the most conducive strategy to our situation.For the strategy layer of the decision-making subsystem, a fuzzy Petri net model is established for the policy option and new methods and ideas is provided for the behavior collaboration of the soccer robot.Third, in the low-level robot action execution level of the decision-making subsystem, each robot has its own task in the process of running towards the target and collision because of the running will affect the behavior collaboration of each robot. Because of this problem, this paper improves the pre-judgement and the formula of the potential field based on the predecessors and realizes the real-time collision avoidance fo the simulation soccer robot.