B-Human — Humanoid Sports Robotics

B-Human at the RoboCup 2014 German Open.

(Quelle: DFKI GmbH/ Universität Bremen)

Football is exciting — for computer scientists as well. That is because the game is a concrete and complex scenario for the development of autonomous, intelligent robots.

Why football? Football is a team sport, where players need to cooperatively find solutions to reach a common goal. To this end, one needs to react continuously to unknown variables: neither how the opposition will act on the pitch and which tactics they employ in specific situations can be predicted, nor the next pass. This scenario is an ideal training ground for computer scientists, as it contains many of the ingredients driving the research in robotics and artificial intelligence.

The B-Human project is based on a cooperation between the Cyber-Physical Systems department and the Department for Mathematics and Computer Science at the University of Bremen, and addresses teaching as well as research. Here, students are learning software development on the job, guided by researchers from DFKI and the University.

RoboCup — more than a Cup

The team behind the team: the 2013 world champions.

(Source: DFKI GmbH/ Universität Bremen)

All of this takes place within an annual competition, which is organised by the international scientific RoboCup initiative for the advancement of research in artificial intelligence and robotics. The competition format allows to directly compare the different approaches. It is divided into separate leagues, each with a different research focus. The overall goal, the ongoing development of robots interacting with their environment, is shared by the Bremen B-Human team as well.

The Standard Platform League, where B-Human is playing since 2009 and counts amongst the most successful teams worldwide, centres on software development. All teams employ the same hardware, the bipedal NAO robot produced by French company Aldebaran Robotics. Only the software makes the difference. B-Human opens the sources of their software after each world cup, so other teams can make us of it as well.

The Challenges — from Sensors to Decisions

A football robot in action.

(Source: DFKI GmbH/ Universität Bremen)

The challenges of robot football put five research areas in the centre of B-Human's software development efforts:

  • External Perception (Image processing of sensor date, recognition of objects in 3D space and their orientation)
  • Internal Perception (Determining the pose of own body and all limbs in 3d space)
  • Probabilistic Modelling (Modelling the environment, a.o. based on calculation of probabilities)
  • Movement (Running, shooting, standing up, looking)
  • Behaviour (Making decisions based on information from oneself and team mates)

The Ultimate Goal: Transfer to Safe and Dependable Systems

The ultimate goal is to develop autonomous robot systems, which act in a safe and dependable way. These developments can be transferred into many other application areas — from space robotics to search and rescue. One example of technology transfer is the Bremen Ambient Assisted Living Lab of DFKI: the software architecture of intelligent wheel chairs developed there is based on the code of B-Human.

(Source: DFKI GmbH/ Universität Bremen)

International Honours in the Standard Platform League

World Cup: 2009, 2010, 2011, 2013, 2016, 2017, 2. Place (2012, 2015)

German Open: 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017

© DFKI GmbH
last updated 31/07/2017
to top