Research On Global Map Construction For Middle-Size League Robot Soccer Competition System

Middle-size league robot soccer system is a very typical multi-agent distributed control system,it is the integration of various fields of artificial intelligence,robotics, sensors and other disciplines.It mainly studies and deals with the problems that in a complex game environment, how to effectively and immediately gain the game environment information among multiple robots,and according to the real-time information realize the mutual communication and coordination between each other, build the global map,and further through the decision subsystem process and motion control subsystem execute the mission instructions, and finally finished the competition task. It provides a standard experimental platform to promote the developments of multiple robots system,artificial intelligence and other fields,the theory and technology can be used in industry, national defense,medical and the practice field, it can also effectively promote national science and technology and the development of the economy. So the research of robot soccer system is highly challenging and has the very high theoretical research value as well as practical application value.Regarding the RoboCup middle-size league robot soccer system as the research object, this paper mainly studied the two aspects of the local map building and global map building.The main research contents are as follows:1.The robot local map buildingCombined with the characteristics of RoboCup middle-size league robot soccer system, the robot local map building is completed on the robot itself. In the data fusion, we adopt the centralized fusion as the fusion structure, on the game court the environment map data which is gathered by the multiple sensors carried by each robot will be directly transmitted to each robot car notebook for processing,based on related local map construction algorithm through data fusion, it can be quickly and accurately processed for robot local map construction. 2.The global map building of middle-size leagus robot soccer systemAfter each robot has completed its local map building, local map information will be wirelessly transmitted to the outside coach for data fusion.When multiple monocular front vision has gained observations of the soccer, this paper carries on the data fusion by using the fusion algorithm based on mean-variance ratio,when multiple omni-vision has gained observations of the opponent, this paper carries on the data fusion by using the fusion algorithm based on the support, by the combination of the two methods the global map building is completed and this method greatly eliminates the influence of the unreal noise data.The last part of this paper summarizes the main research results,points out the shortcomings of the paper and the further research problems to be solved.