机器人世界杯足球锦标赛中多机器人对目标协同定位算法的改进

以机器人世界杯足球锦标赛（ROBOCUP）中全自主机器人平台为对象，针对传统的基于密度空间聚类方法（DBscAN）在对数据处理的精准度和稳定性上的出现的问题，提出了引入机器人距离阈值的改进DBSCAN算法，以提高多机器人协同定位的准确度．首先通过研究单个机器人自定位和目标定位，建立协同定位信息融合的模型；然后该算法通过引人机器人距离阈值，结合平台实际，即机器人测控距离越近测控的精度越高，将传统基于密度空间的聚类方法中有可能被剔除掉的数据，通过与阈值的对比而决定是否保留，提高融合数据的数据量和准确度，从而解决观测信息误差较大以及融合数据不稳定的问题．实验中，在多机器人获取同一个点的情况下分别使用传统DBSCAN和改进DBNscAN算法对目标点进行数据融合．实验结果表明对比传统DBSCAN，改进后的算法在领域半径EPS变化的情况下，融合数据依然稳定．ROBOCUP全自主机器人平台中使用引入机器人距离阈值判断的改进DBSCAN算法进行协同定位，这让其信息融合在稳定性和精确性方面要高于传统的DBSCAN算法．

Improvements of collaborative localization algorithm of multi-robot on target in ROBOCUP

Armed at the accuracy and stability of data processing by density-based spatial clustering of applications with noise algorithm（DBSCAN）, an improved DBSCAN clustering algorithm was used to raise the accuracy of multi-robot cooperative localization on platform of the Robot World Cup （ROBOCUP） autonomous robot. The information fusion model of cooperative localization was established by studying the self-positioning and targeting of a single-robot. The distance threshold was introduced in the improved DBSCAN algorithm, by combining the actual condition of less distance of observation, the higher accuracy, the data which might be deleted by DBSCAN was kept by comparing the actual distance to the distance threshold. In this way, the size and accuracy rating of data was improved and the error of observation information and instability of data was reduced. In the experiment, the traditional DBSCAN clustering algorithm and improved DBSCAN clustering algorithm were both used in the data fusion of same observation point on robot platform. The experiment results show that the integration of data is stable by the improved algorithm with the EPS changes; the data fusion is more stable and accurate using improved DBSCAN clustering algorithm, in which the robot distance threshold is used on real ROBOCUP autonomous robot platform.