I visited two self hosted talks here at RoHOW. The first discussed ball detection using neural networks and the other discussed behavior engineering. At the moment, neural networks for image recognition and image segmentation are commonly used on the Nao hardware. The Bit Bots from the Robocup Humanoid Kid-Size League presented a more debuggable strategy framework. Their paper is on the roadway to being published.
Later during the day, we arranged a testgame with B-Human on the new field. After falling behind 0:1, we managed a achive 1:1 tie at the end. Our localization and behavior worked surprisingly well with the new dimensions of the penalty area and the position of the penalty spot.