The humanoid robots must bring many skills to participate in the football matches. Motion sequences like walking on two legs or shooting are great challenges for a robot. But also the processing of the image data and interpretation of these are the focus of the development. In addition to the software is always worked again and again to improve the hardware.
The human-like running of robots is not very easy. Thus man needs the big toe round his foot to create a dynamic movement. In the case of robots, this is often attempted with a “tap step”. We are currently working on a modified version of the Walkengine of the DARwin-OP, which creates a stable running motion.
In football, after shooting, shooting is of similar importance and therefore also one of our priorities. Here, not only as wide a range as possible and precise meeting are required, but above all stability. As one of the next goals, we want to reach a university.
Image processing is one of our key areas. Here we are working on optimizing the algorithms and increasing the detection rate. The more robust the image processing, the better the behavior of the humanoids can be controlled.
As a further focus, we are currently dealing with localization, that is, The ability of the robots to determine their own position in the playing field. Self-localization is the first step towards a more abstract artificial intelligence.
since June 2014
Rule changes for 2014 have caused us to build larger robots. For this, we have adapted the Nimbro OP to a size of 85 cm. The aluminum and carbon parts were manufactured as before in the university workshop. We created plastic parts, like the head and the hands, with our 3D printer. In addition to significantly larger batteries are used Intel NUCs with i5 processor.
since June 2012
The mechanical design of this generation is based on the DARwIn-OP. With a sturdy aluminum scaffolding and stronger servos (RX-28 or MX-28), we have a convincing new platform. We have the metal parts made in our university workshop. On a new main board is now an Intel Atom processor is used.
September 2010 – June 2012
In addition to some small improvements, two new degrees of freedom have been added.
June 2009 – September 2010
After we noticed on the GermanOpen 2009 that the robot team around JonnyFirst is too heavy, we have dismantled the PDA and now work in the head with an embedded Linux. Both image processing and artificial intelligence are now running on the processor. With this model, we entered the RoboCup 2009 in Graz.
April 2009 – June 2009
With the improved version of Jonny we participated in the GermanOpen 2009. Here we have reduced the head and also made the legs smaller again. Much has also been done on electronics. So we had only 3 controllers at this time.
February 2009 – April 2009
With the so-called “Huno”, the humanoid construction from the RoboBuilder Kit, we have continued. It was equipped with image processing and PDA and we shot our first goal. With JonnyFirst, our extended version of the RoboBuilder kit, we have shot our qualification video.
End of 2008 – February 2009
We made our first attempts with humanoid robots with the Bioloid Kit. We have nevertheless decided, for reasons of cost and since the wCK servos faster, for the RoboBuilder kit.
Since 2009 we are represented in the Humanoid Kid Size League. The special thing about this league is that the robots have to be very similar to humans. For example, they must run on two legs and have physical-human characteristics. As a further difficulty, only sensors are allowed, which also man has. We use cameras (eyes), gyroscope and accelerometer (balance sense) and some of the other robots also have pressure sensors on the feet.
Other special sensors are not permitted. Furthermore, the robots must also play completely autonomously; Which means that the intelligence must also be in the robot and not on another computer, which in turn turns the robot remotely. The robots are allowed to talk to each other via WLAN to discuss themselves.
In the Humanoid League there are two areas: the Games and the Technical Challenges. In the games, four are played against four football, the rules are modeled on the FIFA rules and are only slightly simplified, so that a game can come about. For this, the robots have to bring a whole lot of abilities to play at all: they must be able to move and shoot, stand up independently and of course find the ball.
The Technical Challenges is about testing and working on things that are not yet in the game. In addition, techniques that are currently available in human football are to be learned and improved. Currently, there are four Technical Challenges. Obstacle Avoidance and Dribbling is about dribbling the ball through obstacles in the goal. In doing so, the abilities are identified as obstacles and a path through the obstacles is to be found and planed.
When throwing in, the ability to lift a ball and throw it is thrown. The goal is to integrate the throw into the game as soon as possible, in order to get closer to human football. The double pass is used to promote the abilities “obstacles”, “recognition of fellow players” and “closer shooting”. Another challenge is the high-kick, where the robot is to reach the goal as high as possible.