Omni-vision based localization for middle-size soccer robot

Self-localization is a vital ability for mobile robot's normal operations.The paper summarizes the vision-based localization approaches for autonomous mobile robots in RoboCup competition,including the self-localization and coordinated multi-robot object localization,and the performance of those localization methods is analyzed and compared.Firstly,the geometric localization approach based on road-mark is introduced,it is simple in theory and has been applied abroad;secondly,the probability-based localization based on Bayesian filter is presented,which can arrive better localization performance by combining multisensor data;then,the coordinated multi-robot object localization technology is studied;finally,the trend of research of the vision-based localization method is discussed.