EASE Symposium on Knowledge Representation for Interaction-engaging Robots 2024

Artificial Intelligence can (almost) solve chess. But robots still struggle moving the pieces across the board. While somewhat exaggerated for the purpose of a catchy intro, the previous comparison has a large kernel of truth. Human intuition as to what is easy has proven no guide as to what is easy to achieve computationally — a state of affairs also known as Moravec’s Paradox. The reasons for it are manifold, but even “just” focusing on computational substitutes for cognition, it is not yet known how to build a mind that can cope autonomously and competently with the real world. How should common sense intuition about physics be captured in such a system? How would it model what it can, or cannot do? Considering such a machine will act in a human world, how can it know what it should, or should not do, and decide accordingly? How can it model other minds, human or otherwise, in order to be a good cooperative partner? Such topics related to human-robot collaboration touch many different disciplines including knowledge representation (KR) and robotics. In order to advance the state of the art, close exchange between researchers in these disciplines is beneficial: Robotic applications provide a challenging testbed for cutting edge ideas in KR, as well as a motivation for related research questions. We, at the Collaborative Research Centre Everyday Activity Science and Engineering (EASE) of the University of Bremen would therefore like to bring together experts in the areas of KR and robotics during the first EASE KR Symposium on Interaction-engaging Robots, to be held in Bremen at the Institute for Artificial Intelligence, 26. – 27. March 2024. The scope of the event includes, but is not limited to the following list of topics that we deem as crucial for human-robot collaboration:

•  theory of mind
•  multi-agent cooperation
•  multi-agent planning
•  task negotiations
•  preferences
•  traits
•  capabilities
•  cognitive models of interaction
•  schematic descriptions of interactions
•  obligation
•  permission

The Symposium is planned as a two-day event. The first day of the event will be dedicated to invited talks by experts from the fields of robotics and knowledge representation. The talks will cover different aspects of object- and human-robot interactions ranging from formalization to implementation in a robotic system. We are in particular interested in discussions about what is needed for a robot to engage in collaborative activities with other agents, and what types of abstractions are suitable to reason about interactions. The second day of the Symposium will be dedicated to working sessions in which visiting and local researchers outline open problems, and define a road map that can guide future research. There will also be the opportunity to talk about possible collaborations among the participants of the event, and to visit the laboratory at the Institute for Artificial Intelligence.