Robots are superior to humans in many areas. Yet there is one thing they cannot do: Master everyday activities effortlessly in the way that small children learn to. The numerous “isolated talents” of the machines – for example, high computing capabilities, precise movements, or great usage of strength – do not come together to form a finely tuned package that can react flexibly to new situation. The “Everyday Activity Science and Engineering” (EASE) Collaborative Research Center at the University of Bremen works on a rounded approach towards skills that are needed for robots in their surroundings and for them to understand their own actions – and in turn make suitable decisions.
The researchers have made comprehensive progress in the first four years of the CRC so that the German Research Foundation has now decided to fund their work for a further four years. For AI-based robotics in Bremen, this is a further milestone on the path to permanently establishing its position in the group of international leaders.
Understanding Vague Instructions and Reflecting on One’s Own Actions
A significant challenge when making robots suitable for everyday implementation is linguistic understanding. Simple instructions such as “Make the bed!” or “Set the table!” require the robot to have a great deal of contextual knowledge so that it can then carry out the task. Robots need to initially acquire this knowledge when they are confronted with tasks that they have never completed in exactly the same form previously. In order to progress efficiently and be deemed trustworthy in the eyes of humans, the machines need to additionally understand and be able to justify their own actions.
All of this requires comprehensive cognitive foundations that have been lacking to date. The CRC EASE combines computer science with other disciplines, such as cognitive sciences, linguistics, and psychology, to close this large gap. Research into human decision-making plays an important role in this.
Cooperations Are an Essential Element of this Research Field
“The extension of the CRC EASE shows the strengths that the University of Bremen has built up at the interface between artificial intelligence and robotics,” emphasizes the President of the University of Bremen, Professor Bernd Scholz-Reiter. “The CRC integrates itself excellently in the university’s research strategy, which is not least down to the good cooperation within the high-profile area ‘Minds, Media, Machines,’ where we bring many interesting skills together.”
Professor Michael Beetz, head of the CRC EASE, also views the extension as a success for the networking of parties in the fields of AI-based robotics. “We have been working intensely with partners from other research institutes for a long time now and were able to involve many of them in EASE at an early stage. Additionally, we are dedicated to open research, as we want to incorporate as many interested people as possible into the EASE research area. The challenges are immense and can only be overcome with international cooperation.”
The Future: Robots that Understand Humans
However, Beetz is looking beyond the borders of the Collaborative Research Center towards further research initiatives in AI-based robotics. “The Collaborative Research Center mainly deals with enabling robots to autonomously carry out their tasks and to turn their acquired knowledge into movement,” he explains. “With this approach, we hope to support ill or handicapped persons in the future so that they can lead self-determined lives for longer. The robot therefore needs a representation of the affected human – thus, what this person needs and can. Only in this way will the robot be able to help in direct interaction. That is a new dimension.”
Prof. Dr. Michael Beetz
CRC EASE Spokesperson
University of Bremen
Phone: +49 421 218-64000
Email: ai-officeprotect me ?!cs.uni-bremenprotect me ?!.de