Observations at the Robocup German Open 2015

The Robocup German Open is an annual event where professionals working on robotics and artificial intelligence showcase their latest innovations, exchange knowledge and ideas, and compete in various ways. Apart from rescue robots navigating complex terrain and providing assistance in simulated environments, such as a home and an industrial warehouse, the soccer match played by teams of robots form the most prominently visible part of the event. The Robocup German Open has taken place since 2001 and has steadily grown in popularity over the years. This year’s edition took place in the city of Magdeburg from April 24 until April 26, 2015. As a researcher coming from a Science and Technology Studies (STS) perspective, there were several things that struck me during this year’s Robocup German Open. The following is part personal reflection on my experiences and part speculative thought on the Robocup German Open and innovation in robotics more generally.

Tinkering stuff

Upon entering the main hall of the event, the gusts of technological innovation blow in one’s face. There are many robots to be found, as well as tables with soldering irons, hard drives, network cables, robot parts, laptops. These items are interspersed with empty bottles of coke and beer, cups that used to hold noodle soup, and other items the participants need to keep working grueling hours, often well into the night. Through a continuous and ongoing process of tinkering, the many teams present attempt to improve both hardware and software of their robots, often on the fly, for example using data harvested from the many sensors with which robots are equipped after a soccer match or simulated rescue operation.

Several Nao robots getting ready for their next match.

The star of the aforementioned soccer matches is the Nao, a humanoid robot produced by Aldebaran Robotics (see picture above), which has been chosen as the standard platform for most of the robot soccer matches at the Robocup events. In the past, large teams of Aldebaran Robotics engineers would attend Robocup events in order to make much-needed repairs to robots that were damaged during soccer matches, which led to the more robust hardware that is currently in use. This process of co-production intertwined producer interests and user demands, and led to the adoption of a technology by a large group of users spearheading innovations in robotics.

Despite this highly visible technical process of tinkering, other roles of the Nao robot at the event show clearly that more anthropocentric approaches to robots are never far away. An example of such an approach is an educational short film produced for the city of Leipzig that shows two Nao robots holding hands, walking through the city and taking pictures of humans in the park. The robots even end up falling in love. In a similar vein, Nao robots featured prominently in a project in the Netherlands, where they were used as robot companions for elderly people that appeared to be able to communicate and function autonomously. However, this could only be achieved because a human operator was sending instructions to the Nao from a nearby room.

These observations are not so much meant to debunk the value a human-controlled Nao has in, for example, elderly care. Rather, I raise these ways of presenting the Nao as an autonomous and likable humanoid robot to suggest that robots have a certain allure that is constructed using a particular mixture of highly anthropomorphic elements and cutting-edge technologies. It is this allure that appears to be mobilized in order to make robotics a more appealing near-future technology.

Still, the appealing and often awe-inspiring capabilities of state of the art robots can be suspended in moments of ‘failure’ – a robot misses the ball, brings its opponent to the ground through what appears to be a head bud, or finds itself unable to walk properly on artificial grass. In the Robocup @home competition, where robots were tested in a home environment to assess their ability to provide assistance to humans, many difficulties arose in simulated scenarios. On one particular occasion a human waving and making gestures was unable to draw the attention of a robot, which of course can be problematic in situations where a robot monitors the health of a human being. Although robots performed admirable tasks, such as autonomous recognition of objects in a room and placing these on a map of a room in which these objects were encountered, the environment in question was rather sterile – more like a laboratory and less like a room one would encounter ‘out there’ in the field.

A rescue robot navigating complex terrain

What is perhaps needed is an approach that places robots that are built using technology presently available in the field. Such a process of ‘radical implementation’ could embed robots in real-life conditions whilst generating data needed to generate a form of robot intelligence better adapted to problems encountered in the world.


Emmy takes a break