RoboCup - World Champion 2006 & 2007

After Deep Blue beat the best chess players in the world, the RoboCup was launched by the AI research community as a new, civilized benchmarking competition. In 2004, the team headed by Martin Riedmiller, Martin Lauer, Roland Hafner, and co-team lead Sascha Lange set themselves the goal of developing the world's best AI and winning the RoboCup 2006 in Germany within two years.

Using creative and novel solutions in the areas of Reinforcement Learning and Deep Learning, a football robot and Artificial Intelligence was created, which in 2006 was fully capable of beating all other teams from Europe, China, Japan, and the United States. The reward was winning two world championships, several European championship titles, and a total of three technical awards honoring the annual best scientific contribution.

By participating in the competition, the team gained experience early on in the robust, real-time integration of Machine Learning and Artificial Intelligence on robots and machines. In the process, valuable insights into the requirements and problems in real environments were gained and solution modules were created, which, according to today's standards, equate to level 4 to 5 autonomy.