Vision-based localization for autonomous mobile robot in RoboCup: a survey

Abstract

The work aims to summarize the vision-based localization approaches for autonomous mobile robots in RoboCup competition,and studies the self-localization and coordinated multi-robot object localization.Firstly,the state of arts and categories of localization techniques were presented.According to whether prior pose or the probabilistic approaches are used and how the representations of working environments are constructed,this paper systematically compared and analyzed various self-localization methods.For coordinated object localization,the static and dynamic tracking methods were investigated.Several key issues related to multi-robot collaborative task,especially visual sensor modeling,image processing and feature matching applied in self-localization were discussed respectively.Finally,some existing problems were pointed out and the technique development trends were given.