» Objectives
» Topics of interest
» Program
» Submission of abstracts
» Organizers
» Contact

Object-Action Complexes

Representations for Grounding Perception by Action and Grounding of Language by Interaction

The general topic of the workshop deals with representations for bridging the gap between sub-symbolic low-level robotics and vision domain and the high-level symbolic AI domain. Traditional psychological and artificial intelligence models of natural cognition studied cognition mainly at the symbolic level. However, there is still a significant gap between high-level symbolic and low-level sensorimotor representations. In this workshop we propose to bring together researchers from both areas to discuss approaches for combining continuous domain approaches with discrete representations. We also present the OAC concept that has been proposed within the European project PACO-PLUS as a mathematical formalization for bridging this representional gap. In particular, this formalization tackles the problem of grounding of perception by action and grounding of language through the interaction.

The aim of the OAC concept is to emphasize the notion that objects and actions are inseparably intertwined and that categories are therefore determined (and also limited) by the action a cognitive agent can perform and by the attributes of the world it can perceive. Entities "things" in the world of a robot (or human) will only become semantically useful objects through the action that the agent can/will perform on them. OACs are proposed as a universal representation enabling efficient planning and execution of purposeful action at all levels of a situated architecture. OACs combine the representational and computational efficiency for purposes of search (the frame problem) of STRIPS rules and the object- and situation-oriented concept of affordance with the logical clarity of the event calculus. Affordance is the relation between a situation, usually including an object of a defined type, and the actions that it allows. While affordances have mostly been analyzed in their purely perceptual aspect, the OAC concept defines them more generally as state-transition functions suited to prediction. Such functions can be used for efficient forward-chaining planning, learning, and execution of actions represented simultaneously at multiple levels in an embodied agent architecture.
Topics of interest
  • Representations of innate, acquired and inferred knowledge
  • Multisensory representations for objects and actions
  • Learning from sensorimotor experience
  • Action grammars
  • Plan generation, recognition and modification
  • Learning through coaching and exploration
  • Communication and interaction
9:00 Welcome and Introduction
Rüdiger Dillmann
Karlsruhe Institute of Technology (KIT), Germany
9:10 Bootstrapping Object and Grasping Knowledge with Object Action Complexes
Norbert Krüger and Justus Piater
University of Southern Denmark
» read abstract
9:40 Grounding Language in Object-Centered Affordance
Mark Steedman
University of Edinburgh
» read abstract
10:10 Affordances: The adventures of an elephant in the land of autonomous robots
Erol Sahin
Middle East Technical University - Ankara, Turkey
10:40 Break
11:20 Model free 3D manipulation-recognition and object-categorization in real time for imitation learning in robots
Florentin Wörgötter
BCCN, Göttingen, Germany
11:50 Neurocomputational models for concept and language grounding
Tom Ziemke
University of Skövde, Sweden
» read abstract
12:20 Action-related Places - Bridging the Gap between Symbolic and Subsymbolic Representation in Mobile Robot Manipulation
Andreas Fedrizzi, Freek Stulp, Michael Beetz
TUM, Germany
» read abstract
12:50 Lunch
14:30 The Systems View on Interactive Online Concept Learning
Christian Goerick
Honda Research Institute Europe GmbH
15:00 Grounded humanoid representations: objects, actions and movements
Gordon Cheng
TUM, Germany
15:30 Motor invariants in action recognition
Giorgio Metta
IIT, Italy
» read abstract
16:00 Towards Action Representation based on Acoustic Packages
Britta Wrede, Lars Schillingmann, Katharina J. Rohlfing
CoR-Lab, Bielefeld University, Germany
16:30 Break
16:45 Exploration and Imitation for the Acquisition of Object-Action Complexes
Tamim Asfour, Karlsruhe Institute of Technology (KIT), Germany
Ales Ude, Jozef Stefan Institute, Slovenia
17:15 On Learning and Using Affordances with Humanoid Robots
José Santos-Victor
Instituto Superior Técnico, Lisbon, Portugal
» read abstract
17:45 Learning action primitives in the object-action space
Volker Krüger, Aalborg University, Denmark
Danica Kragic, KTH, Sweden
18:15 Psychology of the OAC
Saskia van Dantzig, Pascal Haazebroek and Bernhard Hommel
Leiden University, Netherlands
18:45 End
Submission of abstracts
Prospective participants are required to submit a one page abstracts until 9. November 2009. Please send your abstract directly to the Workshop organizers (asfour(at)kit.edu). The abstracts will be published on the workshop webpage.
  • Tamim Asfour, Germany
  • Ales Ude, Slovenia
  • Norbert Krüger, Denmark
  • Justus Piater, Belgium
  • Florentin Wörgötter, Germany
  • Mark Steedman, UK
  • Volker Krüger, Denmark
  • Danica Kragic, Sweden
  • Rüdiger Dillmann, Germany
Tamim Asfour
Karlsruhe Institute of Technology (KIT), Institute for Anthropomatics
Humanoids and Intelligence Systems Lab. IAIM Prof. Dillmann
Adenauerring 2
76131 Karlsruhe

E-mail: asfour(at)kit.edu, Web page