Intelligent Systems Lab Amsterdam

Please see the new ISLA website at: http://isla.science.uva.nl

COGNIRON: a Cognitive Robot Companion

Developing cognitive robots whose "purpose in life" would be to serve humans as assistants or "companions". Such robots would be able to learn new skills and tasks in an active open-ended way and to grow in constant interaction and co-operation with humans.

Research group members

dr. Z. Zivkovic
drs. O. Booij
drs. B. Terwijn
dr. ir. B. Kröse

Conceptual representations of space

The UvA works on a representation of space which can be used to have a natural interaction with humans and which reflects the 'hierarchical' nature of space.

Omnidirectional images from our robot will form a low level map (graph) that can be used for navigation. The links between the images define how to navigate (see video). A graph clustering is used to group the images of the low level map into the higher level representation. Interactions with humans are used to 'name' the clusters.

Local features between images are used to compute the similarity and make a 'link'.

Person following A cognitive robot companion needs to detect, track, recognize people, recognize their speech, gestures, emotions, actions etc. We focused first on tracking humans. The robot should know the positions of the people around it in order to approach them in a socially acceptable way respecting the socially acceptable distances to the person.

Various image cues, geometrical constraints, robot motion data and people motion model are combined in a kernel-based object tracking scheme based on approximate Bayesian methods which is a more general version of the commonly used Kalman filter.

Funding

This project is funded by the European Commission.

Links

This project is part of the COGNIRON project and focusses on the Spatial cognition and multimodal situation awareness (RA5) research area.