After an amazing preliminary round with high wins, no conceded goals and just as great quarterfinal and semi-final games, B-Human was able to win in the RoboCup World Championship final against Nao-Team HTWK from Leipzig. Both teams were playing against each other in the final for the fourth time. In 2018, Team HTWK was able to defeat B-Human for the first time – meaning that an exciting final was on the cards. The team from Leipzig took the lead in the first half with a short pass through the Bremen defense. After several missed chances, B-Human was then able to score the first equalizer five minutes before the final whistle. Previously, the Leipzig team had fouled in the Bremen half. B-Human played the subsequent free kick into the opponent’s half and right in front of the feet of the waiting striker, who directly scored a goal. One of B-Human’s feared long shots took place less than a minute later and ended with the winning goal, bringing the score to 2:1.
B-Human Masters Challenges with Deep Learning
One of the special challenges in Sydney was playing next to large windows with ever changing light penetration, which can lead to both light areas and dark shadows on the playing field. B-Human faced these conditions, which are extremely difficult for machine-based image processing, successfully using innovative deep learning methods, amongst other things. In the past years, the ball recognition was supported using neural networks. This time, a so-called convolutional neural network was employed for the recognition of the robots. This system identifies the positions and the distances of all enclosed NAOs by means of a camera image recorded by a robot. The system not only boasts a high recognition rate but also an extremely high execution rate. Only a few milliseconds are needed for the analysis of an image recorded by a robot. The Bremen team presented the scientific work concerning this at the RoboCup symposium following the world championship.
Frontrunners in the Mixed-Team Competition with “B & B”
B-Human also took part in the mixed-team competition alongside the original championship games. For this competition, two teams join forces and compete against other mixed-teams in a small competition. B-Human joined the team Berlin United from the Humboldt-Universität zu Berlin in Sydney to form the “B & B” team. The team cruised into the finals. There they were once more up against the Leipzig NAOs, who had mixed with the Bembelbots team from the Goethe University Frankfurt, to form “Team Team”. The similarly strong teams gave nothing away and the balanced game ended in 0:0. A winner was only found during the subsequent extra time of penalty kicks: B & B shot the decisive penalty, were able to then hold Team Team’s penalty and thus triumphed.
About B-Human
B-Human is made up of around 15 students from the University of Bremen as well as the supervising scientists, Dr. Thomas Röfer from the DFKI research areas of Cyber-Physical Systems, which is led by Prof. Dr. Rolf Drechsler, and Dr. Tim Laue from the University of Bremen. The company CONTACT Software, the leading supplier of solutions for product processes and the digital transformation, has been the main sponsor of B-Human since 2017. Further sponsors are: University of Bremen Alumni Network, igus® and neuland - Büro für Informatik.
Further information:
Official RoboCup World Cup 2019 website:
https://2019.robocup.org/index.php
B-Human team website:
https://www.b-human.de
This link gives you access to photo material, which you can use when copyright is properly cited:
https://b-log.b-human.de
Contact:
Dr. Tim Laue
Faculty of Mathematics and Computer Science
University of Bremen
Phone: +49 421 218-64209
Email: tlaueprotect me ?!uni-bremenprotect me ?!.de
Dr. Thomas Röfer
Deutsches Forschungszentrum für Künstliche Intelligenz GmbH (DFKI)
(German Research Center for Artificial Intelligence)
Research Area Cyber-Physical Systems
Phone: +49 421 218-64200
Email: Thomas.Roeferprotect me ?!dfkiprotect me ?!.de
