Detecting Features of Middle Size Soccer Field using Omnidirectional Camera for Robot Soccer ERSOW

摘要

ERSOW (EEPIS Robot Soccer on Wheeled) is robot soccer developed by Politeknik Elektronika Negeri Surabaya that is designed and implemented on a Middle Size League division by following the rules of RoboCup, an international robot competition. One of the most famous division is a soccer robot, that is divided into two divisions: (1) SSL (Small Size League) and (2) MSL (Middle Size League). There are many research fields related to soccer robot which must be developed in robot ERSOW such as Artificial Intelligence (AI), Computer Vision, Embedded System, Mechanic Systems, and Hardware. This paper focuses on computer vision research for robot ERSOW, especially detecting features of the middle size soccer field, so that specific features of the field like X-junction, T-junction and L-junction can be detected to help robot positioning task where the result is represented into x and y in real-world coordinate. By knowing the position of the features, the robot position can be calculated. The localization system at robot ERSOW uses odometry, which has a large percentage of data errors. Therefore, we attempt to extract the feature of X-junction that is done to find its x and y coordinates and then the obtained coordinate can be used as a reference for correcting odometry data by AI.