Feelix Growing Portal

Calinon, S. and Billard, A. (2008)
Learning Algorithms and Systems Laboratory (LASA), Ecole Polytechnique Federale de Lausanne (EPFL), IEEE/RSJ Intl Conf. on Intelligent Robots and Systems (IROS).

Abstract: We present a probabilistic architecture for solving generically
the problem of extracting the task constraints through a
Programming by Demonstration (PbD) framework and for generalizing
the acquired knowledge to various situations. We propose an
approach based on Gaussian Mixture Regression (GMR) to find
automatically a controller for the robot reproducing the essential
characteristics of the skill by handling simultaneously
constraints in joint space and in task space. Experiments with two
5-DOFs Katana robots are then presented with two manipulation
tasks consisting of handling and displacing a set of objects.

Contact details of the coordinator for queries about the project:
Dr. Lola Cañamero
School of Computer Science
University of Hertfordshire
College Lane
Hatfield, Herts, AL10 9AB, UK
email: L DOT Canamero AT [REMOVE THIS] herts.ac.uk
web page

We are interested in knowing how a robot head can learn to recognize facial expressions without supervision. Our starting point is a mathematical model
showing that a sensory-motor architecture is able to express its emotions succeedes to recognize on-line the facial expression of a caregiver if this latter naturally tends to imitate or to resonate with the system. Interestingly, our works also show that, learning autonomously to recognize a face/non face is more complex than to recognize a facial expression. We propose an architecture using the interaction rhythm to allow first a robust learning of the facial expression without a face tracking and next to perform the learning of the face/non face recognition. Finally we emphasize the importance of the emotions as a mechanism to ensure the dynamical coupling between individuals allowing to learn more and more complex tasks.

This paper discusses the role of two antagonist neural networks for the learning and control of complex behaviors composed as a sequence of elementary states. Learning a pathway with a mobile robot or a sequence of actions with a robot arm can be seen either as the result of the learning of a temporal sequence or as the result of the natural dynamics of a sensory-motor system using appearance based approaches for instance. As a result, we will discuss the performances and the complementary features of each system, and propose a unique control architecture embedding both systems for long life learning.

Submitted to EPIROB 2008

Calinon, S. and Billard, A. (2008)
Learning Algorithms and Systems Laboratory (LASA), Ecole Polytechnique Federale de Lausanne (EPFL)
IEEE Intl Conf. on Robotics and Automation (ICRA), Workshop on Social Interaction with Intelligent Indoor Robots.

Abstract: Bringing robots as collaborative partners into homes presents various challenges to human-robot interaction. Robots will need to interact with untrained users in environments that are originally designed for humans. Compared to their industrial homologous form, humanoid robots can not be preprogrammed with an initial set of behaviours. They should adapt their skills to a huge range of possible tasks without needing to change the environments and tools to fit their needs. The rise of these humanoids implies an inherent social dimension to this technology, where the end-users should be able to teach new skills to these robots in an intuitive manner, relying only on their experience in teaching new skills to other human partners. Our research aims at designing a generic Robot Programming by Demonstration (RPD) framework based on a probabilistic representation of the task constraints, which allows to integrate information from cross-situational statistics and from various social cues such as joint attention or vocal intonation. This paper presents our ongoing research towards bringing user-friendly human-robot teaching systems that would speed up the skill transfer process.