Laboratoire d'Informatique de Grenoble (UMR 5217)
Université Grenoble Alpes
Bâtiment IMAG – 700 avenue Centrale
Domaine Universitaire – 38401 St Martin d'Hères
sylvie.pesty@imag.fr
Office: 365
Phone: (+33) 4 76 51 46 26

A major objective in the area of information technologies is to develop interactive devices that are more attractive and closer to the users and that can be considered as believable interlocutors. This leads to an increasing interest in research works on Artificial Intelligence and Affective Computing, with particular attention to Social Robots, Virtual Characters and more generally to Artificial Companions. Artificial companions can be defined as socially, affective and interactive believable artificial interlocutors with which the user builds a kind of life-long relationship.

Considering such issues, we intended to contribute both at the theoretical level and the application level. Thus, we have investigated the relations between emotions and communication and proposed a Multimodal Conversational Language (MCL) that allows an artificial companion to express in a expressive manner, verbal and non-verbal, and to interact in a more human-like manner. We have also defined a Cognitive and Affective Architecture (BDI-like reasoning approach integrating Emotion) to deduce the communicative intention of an artificial companion from its mental states.

Our main current concern is the way to personalise robot’s behaviour in a particular context. We aim to contribute on both questions of personalisation and context adaptation of companion robots. We invest the notion of style which is a psychological concept that corresponds to the mannerism used to play a particular role in a particular social context.

We implement our technologies in differents social robots and virtual characters :

• Reeti, from Robopec : an expressive head,
• Nao, from Aldebaran, a small humanoid robot, 58 cm high,
• Greta, a SAIBA-compliant ECA from Telecom ParisTech,
• Mary, a virtual character from the M.A.R.C. platform, LIMSI (Paris-Orsay).

We used them in experiments aimed at assessing the benefits of our approaches in terms of perceived sincerity, believability and acceptability of a companion interacting with a human user. Our main issue is social interaction between humans and artificial companions in daily life situations (guarding, playing, assistance, coaching, comforting, teaching,…)

The seminal projet of this research topic started in 2000 and was in the context of the “2nd World”, the Canal+ french virtual city of Paris and a social community of virtual characters (Guillaume Chicoisne's PhD). The work was continued in cooperation with Daniel Vanderveken from the Department of Philosophy, University of Quebec at Trois-Rivières (Quebec) until the end of 2006 (Alexandra Berger’s PhD). It has then been incorporated into the scope of the “PRAIA Project” (Pedagogical Rational and Affective Intelligent Agents), an international project (2007-2010), supported by the Capes-Cofecub program (Michelle Leonhardt’s PhD). It was then part of the ANR CECIL (Complex Emotions in Communication, Interaction and Language) from 2008 to 2012 (Jérémy Rivière’s PhD).

This research is actually mainly conducted within the ANR MoCa project : My little artificial companions world (Wafa Johal’s PhD).

This project focuses on artificial companions for children of ages between 8 and 12 in a home environment. The overall goal is to construct a mixed agent society, composed of companion agents, such as robots and virtual characters, as well as humans. In this society, human-agent relationships will take the same form as with human-human relationships.

This project aims at providing some answers to questions as: Is a group of companions more desirable than a single companion? Can the group help clarifying the social role of each companion, and thus making these companions more acceptable? Furthermore, previous studies do not address the issue of personality. Do users want companions with personality? Can personality make social roles like caring more foreseeable for users? Inside a group of companions, should individual companions have similar or complementary personalities? Do users expect companions with different roles in the group to display different personalities?

We also aim to design the robot companions so as to suit the children’s needs but with a mannerism that is compatible with the parents style of parenting. We want to know if style can explain interpersonal variability in acceptance of social robots. Because of these variabilities, acceptance of robots is considered difficult in home environments. Personification and personalisation of the companion robot can be key to its acceptability. Keywords: social robotvirtual characterEmotionsocial Human-Machine Interaction