DFKI Workshop Session 1: Artificial Intelligence in Robotics
A DFKI Bremen robotics researcher introduces how DFKI combines AI and robotics for long-term autonomy, with a focus on underwater systems and their real-world uses.
In this talk
- 1:28
DFKI’s robotics mission
DFKI grew from a national AI initiative into a network of research centers, with its Bremen robotics center established in 2006. More than 130 researchers work across mechanical and electrical engineering, computer science, biology, mathematics, and physics. The center aims to build robots that can operate autonomously in real environments for months or years.
- 11:26
AI as an interdisciplinary field
Robotics draws on engineering and natural sciences, including cognitive science, linguistics, and neuroscience. The presenter connects this view to Alan Turing’s 1948 paper on intelligent machinery: robots could develop intelligence through sustained interaction with their surroundings and learning from the results. This approach calls for architectures and control methods that can improve through experience.
- 26:01
Learning to handle complexity
A robot senses its environment, chooses actions, and receives feedback about their results. The challenge grows with the robot’s physical structure: more degrees of freedom can make control much harder, especially for closed or parallel kinematic chains. The team uses learning methods and repeated real-world trials to build systems that cope with that complexity.
- 36:59
Autonomous underwater operations
DFKI’s underwater work includes navigation, inspection, and manipulation. FlatFish is designed to live subsea in a garage, leave on missions such as inspecting pipelines, create 3D views of infrastructure, and return to dock. Other systems test deep-sea manipulation and robot-to-robot docking, while the Cuttlefish combines swimming with two manipulators.
- 43:37
Testing on Earth and beyond
Researchers test robots in an indoor basin and at a field site on Heligoland, where teams can work for a week or two. The same capabilities could support exploration of Europa and Enceladus, where robots would need to pass through ice possibly 10 to 20 kilometers thick before exploring the ocean below. Limited underwater communication also makes autonomy essential.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.