| Nowadays, the research of multi-agent system is an important issue and developing direction of the artificial intelligence field. Because of the characters of complexity, non-linear, dynamic and uncertainty of the competition environment and its goal that needs multi-agent cooperation to win the game, the robot soccer becomes an experiment platform for the research of intelligent control and artificial intelligence. A brief introduction about robot soccer system including main technology, history and significance are presented in this paper as well as the theory of robot soccer system, hardware and control system. The main achievements are as follows: 1,Study and design of the three layer control model. Some developments were made in the following fields: (1) Basic Movement Layer: Basic movements of the robot including turning to the appointed angle, getting to the appointed point and moving with appointed speed were designed according to the soccer model of the robot. (2) Path Planning Layer: Research and realization of perpendicular bisector method and potential field method of classical methods. (3) Harmonizer Layer: Control model was set up using area dividing method. 7 kinds of match tactics were set up as well as models that roles assigned using the relative position method. 2,A kind of control model consists of two layers was set up. Robot was regard as an intellectual body with some intelligence other than merely an executive body. Using the advantage of centralized control and multi-agent plan theory, a framework model including one control coordinate intellectual body and 3 plan carry out intellectual body was build up. Relevant programs were designed too. 3,A kind of shoot algorithm was designed utilizing reactive agent theory. External environment was regard as a kind of stimulation, the robot does not need to make path plan and just to act by this stimulation. The research has tested and experimented on the robot soccer program owned by BJUT.
|
PDF Full Text Request