Learning Odors for Social Robots: The URBANO Experience

Daniel Galán, Ramón Galán, Ángel Luis Martínez, Iveth Moreno
unpublished
This paper presents the experience of the Intelligence Control Group of UPM in the design of URBANO, a tour guide robot. It is a cognitive system based on distributed agents. One of these agents is an ontology that contains the knowledge used by the robot. This knowledge is mainly developed for linguistic applications. Here it is described how to add odor experiences to some available concepts in the ontology. Odor experiences evolve in time so the learning process must be adaptive and
more » ... d. Neural networks, fuzzy logic, recursive least squares, Mahalanobis distance and genetic algorithms are tested over a low-cost multipurpose electronic nose in the URBANO environment. The obtained results show how to add odors to the emotional model of the robot help it to increase its performance as a social robot.
fatcat:lakst4hli5aljfb3iok2adqka4