Real-time self-localization of a mobile robot by vision and motion system

Abstract

In this study, an autonomous robot is designed to equip an omni-vision camera and an omni-moving platform to meet the requirement of FIRA and RoboCup robot soccer competitions. To make robots obtain its location on the field, we use white-line pattern match localization algorithm to achieve this goal. However, when the white-line information is not complete during the matching process, the observed data has a huge difference with the pre-build database. Then the localization result will cause errors. In this study, we first introduce encoder localization algorithm to get robot's moving direction and even trace its motion as an odometer. Then we integrate the white-line pattern match localization and encoder localization as the proposed algorithm. The results demonstrate that the inetgration localization outperforms than the localization with white-line pattern match localization or encoder localization only. We can obtain robot location within 40 ms and with an error less than 10 cm by the algorithm. Meanwhile, the integration localization algorithm is compared with other schemes to show its accuracy.