CogX: Cognitive Systems that Self-Understand and Self-Extend
Dora the explorer robot. Image: CogX Project Website.CogX: Cognitive Systems that Self-Understand and Self-Extend
CogX was an EU FP7 ICT project that ran from May 2008 to June 2012 and brought together six research groups - University of Birmingham (coordinator), DFKI Saarbrücken, KTH Stockholm, University of Ljubljana, University of Freiburg and TU Wien - to explore how robots can recognize gaps in their own knowledge (“self-understanding”) and decide what to learn next (“self-extension”). The consortium produced a unified architecture that fused probabilistic perception, semantic mapping, manipulation, dialogue and deliberative planning, and demonstrated its ideas on mobile platforms such as Dora, a robot able to explore unfamiliar indoor spaces, reason about its uncertainties and autonomously plan sensing or question-asking actions to resolve them. CogX advanced techniques for uncertainty-aware representation and active learning, providing foundations for more adaptable cognitive robots suited to open-ended, dynamic environments.
Roles
2009 - 2012: Researcher | Ph.D. Student @ FRI
- Developed self-supervised multi-view online learning vector quantization algorithms & integrated work within the overall CogX project software framework.
- Developed feature relevance determination methods for learning vector quantization algorithms.
- Co-author of several deliverables, project reports & presentations.
Videos
George Y2 - Interactive learning in dialogue with a tutor
I am the narrator of the above video.
Video: Cogx1 YouTube Channel. Credit: University of Ljubljana, CogX Project.
