You were recently appointed as a Research Fellow at fortiss. Here you will support the institute in the area of knowledge-based systems, among other things. Can you tell us something about your vision and goals in this position?
My research at the Institute for Artificial Intelligence (IAI) at the University of Bremen focuses on knowledge-based systems that can be used in robotics. My vision includes the realization of robots that know exactly what they are doing, why they are doing it and how they are doing it. This will enable them to predict the consequences of their actions and adapt their actions autonomously to avoid undesirable outcomes. This will make robots more autonomous and intelligent. They will be able to operate effectively in complex, unknown environments.
What experiences have you already had working with fortiss and what strengths do you see in the institute?
The IAI and fortiss have already cooperated on the development of the Knowledge4Retail platform for the retail sector. The platform connects online and stationary retail, supports strategic marketing and enables digital solutions for personalized customer service. I have also worked closely with my fortiss colleagues on other projects, for example when I was spokesperson for the CoTeSys (Cognition for Technical Systems) Cluster of Excellence at the Technical University of Munich. In the EU project RoboEarth, we conducted joint research into cloud-based knowledge services for robotics.
The strengths of fortiss lie in the close networking of research and practice as well as the collaborative way of working within and outside the institute. I am particularly impressed by the institute's ability to effectively transfer research results to industry.
Your research at the Institute for IAI focuses on AI-based control methods for robots, particularly with regard to everyday manipulation tasks. How do you see the transfer of these research results to the area of platform engineering at fortiss?
My research focuses on robotic tasks in everyday scenarios, which are often vaguely formulated. This means that they are expressed colloquially and require a great deal of background knowledge that we humans take for granted, but which robots do not readily possess. This becomes relevant when the industrial robots leave their assigned workstations and become direct work colleagues of humans in the production facilities.
Platform engineering, which fortiss is driving forward, should enable high-tech companies in particular to develop and implement innovative solutions more efficiently. The transfer of my research results can help to develop more flexible and adaptable platforms that are tailored to different requirements and environments. This flexibility facilitates the development and integration of advanced robotic solutions. This significantly reduces the barriers to their use in companies.
As coordinator of the German Collaborative Research Center EASE (Everyday Activity Science and Engineering), you are working intensively on giving robots human-like abilities to perform everyday tasks. What role does this research play in the development of future platforms and technologies?
In the Collaborative Research Center EASE (Everyday Activity Science and Engineering), we are researching how everyday tasks are performed by humans and, on this basis, developing the necessary skills in robots so that they can also perform these tasks efficiently. To this end, we analyze and model the cognitive foundations of human action in order to reproduce them in AI systems. In this way, we can develop robots that can perform a wide range of tasks in different environments - from domestic settings to industrial applications.
By developing robots with cognitive capabilities, we are paving the way for the creation of versatile and autonomous robotic platforms. These platforms need to improve their capabilities and autonomy to meet complex challenges. Our vision is to better align these requirements so that robotic systems can adapt to different and dynamic environments and prove themselves in the field
Robots that have a better understanding of their tasks and better implementation skills can lead to significant innovations in various areas. In healthcare, for example, they could help with patient care and rehabilitation exercises. In manufacturing, they could take over complex assembly tasks. Ultimately, our research aims to create a foundation for robots that are not just tools, but competent partners to humans, significantly improving quality of life and productivity.
Your work on openEASE aims to improve interoperability in robotics and lower the barriers to robot programming. How could these efforts influence the development of platforms for software-intensive systems at fortiss?
openEASE serves as an open knowledge service that provides environment models and robot models together with comprehensive knowledge databases. It is part of the open research portfolio in Bremen, which also includes numerous other services. We think it is very important to make research tools openly accessible. We see great potential in our Virtual Research Building, which not only makes our own services easily and clearly accessible, but also those of other research institutions. For example, we make our robotics labs available online so that scientists and developers worldwide have access to our robots and environments. In turn, they can then make their results available to the entire research community.
By promoting open collaboration and resource sharing, openEASE enhances our ability to advance cognitive robotics and develop smarter, more powerful systems. Due to this extensive expertise, the IAI is also one of the coordinators of the European robotics and AI network euROBIN, which brings together leading research institutions and companies.
With our activities, we can significantly support the development of platforms for software-intensive systems at fortiss by facilitating access to useful tools and promoting collaboration between different players.
You received the ERC Advanced Grant from the EU last year to support basic research in the field of anticipatory action. Can you tell us more about this project and its potential importance for the future development of AI and robotics?
In the European Research Council (ERC) funded Advanced Grant "FAME" (Future-oriented Cognitive Action Modeling Engine), our work focuses on understanding and developing the cognitive capabilities necessary for robots to predict the outcomes of their actions. ERC funding supports both basic and applied research and innovation transfer. The grant will deepen our understanding of how robots can represent planned activities internally and put them into practice. These predictive capabilities will enable robots to perform more complex tasks in unfamiliar environments.
A key aspect of this research is the integration with basic research in the Collaborative Research Center EASE. Here we are investigating how robots can use internal simulations to visualize and understand their environment and actions. For example, our FAME robots have a game engine system that enables them to simulate their environment and plan their actions accordingly. This capability allows them to see potential plans in their "mind's eye", understand the implications of their actions and refine their behavior based on these insights.
The potential significance of this research for the future development of AI and robotics is immense. By improving the cognitive and predictive capabilities of robots, we are paving the way for more autonomous, intelligent and adaptive robotic systems that can explain their own actions rather than acting as a 'black box'. These advances will be crucial in various fields, from industrial automation to healthcare, where robots need to work in dynamic and unpredictable environments. Ultimately, our work aims to develop robots that can plan ahead, make informed decisions and integrate seamlessly into human-centered environments, which will significantly increase their utility and impact.
With the Knowledge4Retail project, in which the University of Bremen and fortiss were project partners, you want to establish a new generation of information systems for retail and its supply chains. How could these innovations enrich the area of platform engineering at fortiss?
The Knowledge4Retail project has set itself the goal of strengthening the retail sector through the use of AI and robotics. A key innovation is the use of realistic digital twins of retail stores. These virtual models allow us to use robots and build automated systems to optimize warehousing and inventory management. Such applications are typical of digital information platforms and will enable us to support platform development at fortiss with smarter, more responsive and efficient systems.
What other projects will you be supporting or driving forward at fortiss?
One major project is the KI.Fabrik Bayern, where I see great synergy effects between the IAI and fortiss. For example, I can imagine a cooperation to make the AI.factory in our virtual research building accessible to interested parties from all over the world.
We also want to integrate ontology- and knowledge-based modeling of robot functions in industrial environments. This is crucial for our research because it gives us better access to innovation and transfer. In this way, we improve the social and economic impact of our work and ensure that the latest advances in AI and robotics are used effectively and beneficially in different industries.
Your vision is to develop AI methods that enable robots to perform vaguely formulated everyday tasks independently. What role does this vision play in shaping the future of platforms and systems?
In the context of generative AI, the ability to cope with vaguely formulated tasks such as "Set the table!" considerably expands the scope of application of AI. However, it also poses a scientific challenge because AI often calculates with probabilities based on its existing knowledge without guaranteeing accuracy. Therefore, it is essential to link generative technologies with knowledge-based systems to make AI more transparent and trustworthy. This integration will help to ensure that the decisions and actions of AI are understandable and reliable. This is fundamental to the development of future platforms and systems.
fortiss is an application-oriented research institute and thus a bridge between science and industry. Are there any other transfer offers that you would like to develop for companies with fortiss?
Our research is very application-oriented and is based on the principles of "open research" to facilitate collaborations. Examples of this include the aforementioned Knowledge4Retail project, which has won several science and transfer awards, and its spin-off Ubica Robotics, which offers robots for the retail sector. We also cooperate with companies to implement control programs for knowledge processing in robots. The input from fortiss enables us to improve these programs and make them more robust and useful for industrial applications. This approach will help companies integrate advanced AI and robotics technologies, drive innovation and increase operational efficiency.
