Abstracts

  • Tamim Asfour, KIT
    Engineering 24/7 Humanoids with Functional Intelligence
    Research in humanoid robotics strives to create versatile machines endowed with functional intelligence, i.e. the abilities to reason and act in the physical world and to perform any task that a human might reasonably be expected to carry out. The era of humanoid robots acting in the real world as general-purpose machines with functional intelligence is dawning. Substantial advances have been made, positioning humanoid robotics as a cornerstone in both robotics research and understanding intelligence. In this talk, I will discuss progress towards 24/7 humanoid robots for empowering and assisting humans in daily life. I will specifically focus on learning from human and natural interaction, on performing complex manipulation tasks and on engineering holistic systems with architectures integrating semantic representations and sensorimotor control. I will conclude with a discussion of the major challenges and the transformative impact humanoid robotics holds for the future.

  • Michael Beetz, Universität Bremen
    Empowering Robots with Digital Mental Models: Filling the Cognitive Gap for Everyday Tasks
    In this talk I introduce Digital Mental Models (DMMs) as a novel cognitive capability of AI-powered and cognition-enabled robots. By combining digital twin technology with symbolic knowledge representation and embodying this combination into robots, we tackle the challenge of converting vague task requests into specific robot actions, that is robot motions that cause desired physical effects and avoid unwanted side effects. This breakthrough enables robots to perform everyday manipulation tasks with an unprecedented level of context-sensitivity, foresight, generality, and transferability. DMMs narrow the cognitive divide currently existing in robotics by equipping robots with a profound understanding of the physical world and how it works.
  • Stefano Borgo, Istituto di Scienze e Tecnologie della Cognizione
    Places and interaction places
     We focus on interaction as a relationship, among (physical) agents and objects, that develops in time. First, we look at how we can differentiate agents and at how types of interactions can emerge depending on the agents’ characteristics.
    After giving due room to the fact that an agent’s behaviour in an interaction is driven by his/her/its purposes and the available resources, we turn out attention on the very place where the interaction happens and look at it as a driver, and perhaps even a starter, of interactions. The influence of the environment in interactions is well known for human and social agents in general, but is less studied in more general settings, and in particular when interaction involves artificial agents. Overall, our analysis provides a new twist on the study of interaction because it is based on applied ontology methodologies, and because it aims to understand places as carriers of capacities to facilitate, push or hinder different (types of) interactions across heterogenous agents.

  • Maria Hedblom, Jönköping University  
    Pieces of Mind: The cognitive components of thought and reasoning and their formal correspondence 
    Some of the main cognitive theories for human conceptualisation and reasoning propose that there are a finite number of cognitive patterns laying the foundation for meaning and symbol grounding. This talk will present some perspectives on these cognitive patterns and culminate in the theory of image schemas; a particular form of conceptual micro patterns learned from embodied experiences. Often described as spatiotemporal relationships, image schemas are used in a range of cognitive processes, e,g. linguistic expressions, metaphoric and abstract conceptualisation, and analogical and commonsense reasoning. Demonstrated as central components in reasoning and communication, their potential to improve “intelligence” in computational systems is high. Yet the unavoidable problem is: how does one turn an abstract cognitive phenomenon into a machine-readable format? Addressing this issue, some of our previous work on turning image schemas into machine-readable formats will be presented as an inspirational starting point.

  • Stefan Kopp, Universität Bielefeld
    Interaction-driven mentalizing in robots 
    The ability to understand the mental states of other agents is essential for any kind of collaborative activity. This also holds for robots that are supposed to interact with humans in shared environments such that computational models for „mentalizing“ about the user, i.e. representing, inferring, and predicting her relevant mental states, are needed. However, mentalizing is an inherently uncertain and complex inverse planning problem, usually treated as an offline inference problem to which different approaches exist, from logics-based, to probabilistic, to using machine learning and recently using LLMs. When we want to develop robots that can engage in live interaction, however, we need them to reason about another agent’s mental states and to decide upon own actions in a fast and mutually adaptive manner. I will present cognitively inspired work towards „interaction-driven mentalizing“ that aims to allow for computational accounts of thinking about others and interacting with them at the same time.

  • Alessandra Sciutti, Istituto Italiano di Tecnologia
    Cognitive Robotics for Effective Human-Robot Interaction
    Creating robots capable of meaningful interactions with humans requires endowing them with the ability to establish and maintain a nuanced understanding of human cognition. This involves equipping robots with the capability to grasp human needs, desires, and intentions, coupled with the demonstration of clear and intelligible behavior. The discussion extends to the importance of enabling robots to anticipate the consequences of their actions and those of humans, adapt to dynamic scenarios, and autonomously choose appropriate responses. By drawing insights from human cognitive development, we explore the integration of these skills into the design of cognitive robots, significantly amplifying their effectiveness in interacting with humans.