University of Bremen: DFG funds SFB EASE for a further four years
Robots are already superior to humans in many areas, but there is one thing they cannot do: find their way around effortlessly in everyday life, as even small children learn. The numerous “island talents” of the machines – for example high computing power, precise movements or large amounts of force – do not come together in a finely tuned overall package that can react flexibly to new situations. The Collaborative Research Center “Everyday Activity Science and Engineering” (EASE) at the University of Bremen takes a holistic approach to the skills required so that robots can understand their environment and their own actions – and make the right decisions based on them.
The scientists made extensive progress in the first four years of the SFB, so that the German Research Foundation has now decided to support the work for a further four years. For AI-based robotics research at the Bremen location, this is another milestone on the way to permanent establishment in the top international group.
Understand vague instructions and reflect on your own actions
An essential hurdle on the way to everyday suitability of robots is the understanding of language: Simple instructions like “Make the bed!” Or “Set the table!” Require a lot of contextual knowledge so that they can be carried out correctly. Robots only need to acquire this knowledge when they are confronted with tasks that they have not already mastered in exactly the same form. In order to be able to proceed efficiently and to be trustworthy for humans, the machines must also be able to understand and justify their own actions.
All of this requires extensive cognitive foundations that were previously lacking. The SFB EASE combines computer science with other disciplines such as cognitive science, linguistics and psychology in order to close these large gaps. Research into human decision-making plays a very important role in this.
Cooperations are an essential part of the research field
“The extension of the SFB EASE shows what strength the University of Bremen has built up at the interface between artificial intelligence and robotics,” emphasizes the Rector of the University of Bremen, Professor Bernd Scholz-Reiter. “The SFB fits in perfectly with the research strategy of the university and is due not least to the good cooperation in the scientific focus ‘Minds, Media, Machines’, in which we bring together many exciting competencies.”
Professor Michael Beetz, head of the SFB EASE, also sees the extension as a success of the increasing networking of actors from AI-based robotics. “We have been working closely with partners from other research institutions for a long time and have integrated many of them into EASE at an early stage. In addition, we are strongly committed to open research because we want to involve as many interested parties as possible in the EASE research field. The challenges are great and can only be mastered in international cooperation. “
Future: robots that understand people
Beetz is already looking beyond the Collaborative Research Center to other research initiatives in AI-based robotics. “The main aim of the Collaborative Research Center is to enable robots to carry out their tasks independently and to convert the knowledge they have acquired into movements,” he explains. “With this approach, we would like to support sick or handicapped people in everyday activities so that they can live longer independently. The robot then also needs a representation of humans – what they want and can do. This is the only way he can help him in direct interaction. This is a new dimension. “