The @Home League of the RoboCup is the largest annual competition for assistance robots to support people in need of help in everyday environments. In the open class, in which teams can design their own robots, 17 teams from eleven countries competed at the world championship in Eindhoven. The robots demonstrated their skills in ten tests in realistic domestic environments: for example, the robots had to welcome visitors and introduce them to other people, help carry luggage, enforce house rules, put away the shopping, tidy the kitchen, set the breakfast table and serve guests in a restaurant. The robots thus proved that they can interact with users using speech and body language, navigate autonomously around obstacles and pick up, transport and place everyday objects.
Two tests in which the task to be solved was communicated by voice commands were particularly challenging. Team NimbRo used two mobile robots with a human-like upper body and omnidirectional drive for these tasks. These robots perceive their surroundings using cameras, laser scanners and a microphone and control numerous motors and a loudspeaker to perform assistance tasks autonomously. Methods from artificial intelligence research, such as image and speech understanding, action and movement planning and dialog systems, are crucial here.
In the preliminary round, the team from the University of Bonn was set back by a hardware defect in the main robot and finished in second place, behind last year's winner Tidyboy from Korea. NimbRo then overtook Tidyboy in the main round. In the final, the three best teams performed self-defined tasks on the topic of “Helping to prepare dinner”. Team NimbRo used two robots at the same time. These demonstrated to the jury the recognition of objects without restriction to predefined categories, the recognition of gestures and interaction with users through a speech dialog system, mobility in an apartment, grasping objects and the integration of multimodal base models for creating recipe suggestions based on recognized ingredients. For this demonstration, NimbRo received the highest score from both the @Home League judges and the industry judges. NimbRo won the competition with 8,852 points, ahead of Tidyboy (7,495 points) and SocRob@Home (Portugal, 6,901 points).
“In the future, assistance robots will contribute to enable people in need of help to live independently in their familiar surroundings for longer,” says Prof. Dr. Sven Behnke, head of the Autonomous Intelligent Systems chair and director of the Institute of Computer Science VI – Intelligent Systems and Robotics at the University of Bonn. “The fact that our team prevailed against strong international competition underlines the excellence of robotics research in Bonn,” emphasizes the scientist, who is also a member of the Transdisciplinary Research Areas (TRAs) ‘Modelling’ and ‘Sustainable Futures’, the PhenoRob Cluster of Excellence and the Center for Robotics at the University of Bonn and the Lamarr Institute for Machine Learning and Artificial Intelligence.
