Social interaction for object recognition and tracking.
Résumé
The paper presents an experiment aiming at studying how an autonomous robot can learn to recognize and classify objects through human-robot interaction. In this experiment, a minimal architecture merging navigation, object recognition, visual tracking, emotional values attribution and human-robot interaction is tested at the Quai Branly museum for 10 days and 4 hours a day. In this experiment, the capability of our architecture to control a mobile robot in a crowded environment with fragile artworks has been validated.