Organisation: Rheinische Friedrich-Wilhelms-Universität Bonn, Autonomous Intelligent Systems
Address: Friedrich-Ebert-Allee 144, 53113 Bonn, Germany
Contact: Prof. Dr. Sven Behnke (Tel: +49 228 73 4422, email@example.com
Team NimbRo Manufacturing is based in the Autonomous Intelligent Systems Group at the Computer Science Institute of University of Bonn, Germany.
The University of Bonn is one of the leading public research universities of Germany. Its Computer Science Institute has a long tradition in robotics research, especially in the areas of mobile robotics and SLAM. Since 2008, the Autonomous Intelligent Systems (AIS) group conducts research in cognitive robotics and machine learning.
The AIS group has extensive experience in real-time perception, control, and system integration for complex robots, including cognitive service robots capable of mobile manipulation and intuitive human-robot interaction in complex environments. We developed leading RGB-D SLAM, semantic mapping, perception of persons, and planning methods for navigation and manipulation.
The AIS group has extensive experience in robot competitions and challenges. Our team NimbRo participates with great success at the international RoboCup competitions, winning a total of 13 tournaments, including the last five years of the Humanoid TeenSize soccer league and the last three years of the @Home domestic service robot league. Recently, we developed also a robot for mobile manipulation in rough terrain, which participated at the DLR SpaceBot Cup.
Our team has the necessary expertise to address the objectives of Challenge1.
We developed the cognitive service robots Dynamaid and Cosero which have an anthropomorphic upper body and an omnidirectional drive. These robots are equipped with multiple laser scanners, an RGB-D camera, a high-resolution camera, a directed microphone, an Intel Core-i7 Quad onboard PC, and a high-performance wireless module. We developed efficient methods for environment mapping, object detection, perception of persons, gesture recognition, navigation and manipulation planning. In the RoboCup@Home competitions, the robots demonstrated picking and placing objects, dual-arm manipulation of larger objects, the opening and closing of doors, the use of tools, intuitive multimodal interaction with persons, and many other skills. In the FP7 ECHORD experiment ActReMa, we implemented bin picking for these mobile robots. In the new FP7 project STAMINA, we develop bin picking for the transportable robots UR10 and Baxter.
The team members cover all relevant aspects, from work space mapping, over object detection and recognition, perception of persons and gesture recognition, to manipulation planning. Our team has not only the necessary technical expertise, but also plenty of experience with robot competitions and challenges. We also have experience in the cooperation with project partners from industry.
- Bayesian Exploration and Interactive Demonstration in Continuous State MAXQ-Learning (PDF)
- Learning Depth-Sensitive Conditional Random Fields for Semantic Segmentation of RGB-D Images (PDF)
- Active Recognition and Manipulation for Mobile Robot Bin Picking (PDF)
- Multi-Resolution Surfel Maps for Efficient Dense 3D Modeling and Tracking (PDF)
- Depth-Enhanced Hough Forests for Object-Class Detection and Continuous Pose Estimation (PDF)
- Compliant Task-Space Control with Back-Drivable Servo Actuators (PDF)
- Combining Contour and Shape Primitives for Object Detection and Pose Estimation of Prefabricated Parts (PDF)
- Joint Detection and Pose Tracking of Multi-Resolution Surfel Models in RGB-D (PDF)
- Learning to Interpret Pointing Gestures with a Time-of-Flight Camera (PDF)