Bielefeld University Robotics Presents Interaction and Learning at the AUTOMATICA

Bielefeld University, along with its Excellence Cluster Cognitive Interaction Technology and its Research Institute for Cognition and Robotics (CoR-Lab), will present application oriented research results at the AUTOMATICA (May 22-25, 2012) in Munich to an international audience. A broad variety of exhibits – from self-learning robot control to human-robot interaction – will provide insight into frontier research on interactive intelligent systems in hall B3, booth 429/427. A highlight of our exhibition is that it is the first time that the free standing Festo Bionic Handling Assistant robot, based on self-learning through child-like testing, is to be presented.

Kinematic Learning for the Festo Bionic Handling Assistant Robot
The “Bionic Handling Assistant” is a new robotic-platform developed by Festo that allows, through compressed-air drives and light and flexible material, safe, physical interaction with humans. Its innovative design was awarded the Deutsche Zukunftspreis (German Future Prize) in 2010. Automatic handling of this bionic “elephant trunk” poses problems for classical approaches, since its exact behavior can only be roughly described. Bielefeld scientists have developed an approach based on early childhood learning to circumvent some of these issues. Systematic grasping attempts, instead of coincidental exploration, fosters efficient learning and quickly lead to applicable results.

Anthropomorphic Robot Head Flobi
The anthropomorphic robot head “Flobi” was developed at the University of Bielefeld. It allows for congenial-emotional facial expressions to be shown on its face. The goal is to develop anthropomorphic, social robots, which can enable more intuitive human-robot interaction and therefore promote stronger acceptance among humans. The design of the head is modular and features such as the eyes and hair are exchangeable in order to make systematic research possible. At AUTOMATICA the reaction and expressive capabilities of the system will be on show during a live demonstration in which Flobi will replicate the head movements of people via an intertial-sensor. The potential fields of application for Flobi are human-machine interfaces in which social and emotional aspects play a role. These are, in particular, motivation and learning scenarios in which non-verbal feedback can be provided via facial expressions or situations in which attention must be coordinated.

Humanoid Robots Interact with Several Humans
The HUMAVIPS  project, funded by the 7th EU framework programme, has the goal to provide humanoid robots with the necessary perceptive capabilities that can enable them to interact with groups of humans. Results of the integration of algorithms developed for audio-visual fusion, for tracking and recognizing the line of sight of people, and for the identification and classification of people will be presented at the AUTOMATICA 2012. The research and technological development conducted within this project focuses on approaches that can perceive multi-modal data and its integration into models and systems of human-robot interaction. For this purpose, a new head with stereo cameras has been developed for the humanoid robot NAO (Aldebaran Robotics). In addition to this, scientists at Bielefeld have created an improved software architecture platform which facilitates scientific work with this robot. Further contributions have consisted of new methods to adapt dialogue systems to enable them to have open interaction with groups of humans, as well as the development of an integration-architecture platform.

Interaction Space for Human-Robot Cooperation
To enable close cooperation between humans and robots, it is necessary to perceive and coordinate actions on both sides. Working towards this end, the scene is first observed and analyzed through a depth camera and a background model is constructed dynamically, in which objects in the foreground are automatically segmented. The arm of the robot can then grasp the object indicated by the human. The research presented here on the visual 3D analysis and modeling of the interaction space can be applied to many areas of service robotics .

The objective of Bielefeld scientists is to make technical systems - ranging from everyday devices to robots - more intuitive and easier to handle. Two institutions at Bielefeld University, the Excellence Cluster Cognitive Interaction Technology (CITEC), funded within the framework of the Excellence Initiative by the Federal Government and the states since 2007, and the Research Institute for Cognition and Robotics (CoR-Lab), are dedicated to developing cognitive interaction technology platforms and systems to help bridge the gap between humans and machines. Participation in numerous EU research projects, along with research cooperations with partners from Japan and North America, as well as collaborations with industrial partners (for example the Leading Edge Cluster Intelligent Technical Systems OstWestfalenLippe “it’s OWL” project, funded within the framework of the Federal Government’s Hightech Strategy) confirm the competitiveness of the research at Bielefeld in a highly competitive field, which is sure to have significant importance for the future.

More information at:
www.cit.ec.de www.cor-lab.de