Utilizing the Structure of Field Lines for Efficient Soccer Robot Localization

摘要: Abstract Self-localization on the field is one of key perceptual tasks that a soccer robot, e.g. in RoboCup competitions, must solve. This problem becomes harder, as rules more and discourage solely color-based orientation field. While size increases, boundary markers goals become smaller less colorful. For robust game play, robots, therefore, need to maintain probabilistic pose estimate rely subtle environmental clues. Field lines are particularly interesting features, because they hardly ever completely occluded observing them significantly reduces number possible poses In this work, we present method for line-based self-localization Unlike previous our first recovers line structure graph from image. From graph, can then easily derive features such corners. Finally, describe optimizations efficient use derived part...