全自主机器人足球系统的全局地图构建研究

研究和讨论了如何通过多机器人的协作，实现全局地图的构建．在单个机器人通过自身携带的多传感器进行局部地图构建的基础上，研究了前向单目视觉传感器的建模方法，在此观测模型的基础上，用极大似然融合算法对球的位置信息进行融合，而对于多机器人返回的对方机器人位置信息，使用基于密度的空间聚类算法（DBSCAN）进行信息融合，从而实现全局地图构建．实验结果表明，通过多机器人的协作，可以准确地构建出全局地图，弥补了单个机器人自身传感器的有限感知范围，本文研究的方法完全满足全自主机器人足球比赛中动态环境地图构建的需要．

Building global map in autonomous robot soccer system

This paper is concerned with the cooperation between robots in building a global map. Based on the local map-building through multi-sensors carried by a single robot, this paper studies the monocular front-vision modeling method. Meanwhile, we elaborate two information fusion methods: density-based spatial clustering of applications with noise(DBSCAN) clustering algorithm and maximum likelihood estimate fusion algorithm. By these two methods, the information about the location of opponent robots and the information about the location of the ball are fused, thus, a global map is built. Experimental results show that through the cooperation of robots, a global map can be accurately built, which compensates the information limitation in sensors of each single robot, and meets requirements in map-building in the dynamic environment of autonomous robot soccer tournament.