PH.D. SCHOLARSHIP for the duration of 3 years on Adaptative and Ambient Animated Agents

-----------------------------------------------------------
PH.D. SCHOLARSHIP for the duration of 3 years on Adaptative and Ambient Animated Agents
-----------------------------------------------------------

----
CONTEXT

An Animated Agent is a multimodal interface in which an animated character displayed on the screen combines several human-like modalities such as speech, gesture and facial expressions. Using an Animated Agent is expected to lead to an intuitive interaction since they use intuitive modalities. Computational models have been developed over the years for generating non-verbal behavior and for evaluating such agents in different settings.

----
RESEARCH GOAL 

This thesis is part of the European project ATRACO (http://www.uni-ulm.de/in/atraco) and is concerned with the inclusion of an animated agent within an ambient framework (user’s home is equipped with sensors and actuators). The use of an animated agent sounds quite relevant in such an ambient framework as it is expected to enable the adaptation of the system according to several criteria: 
•	For a given communication goal (e.g. pointing to a real-world object), the agent can use more or less redundantly the different modalities (e.g. speech, facial expressions, posture, gaze, head orientation), and be more or less intrusive / implicit depending on the context ; for example it should not disturb verbally the user while he is speaking over the phone but it can use a pointing gesture to draw user’s attention to an urgent issue
•	It can use a specific personality profile (eg extrovert vs. introvert) according to user’s own personality, preference or current mood
•	It can adapt its behavior to the location of the user (thanks to automatic detection of user’s position, the agent can gaze at him at the proper location)
•	It can adapt its appearance according to different gender and age depending on user’s preference (male / female, adult / children)


The goal of the thesis is to model, implement and evaluate an animated agent that enables to adapt the human-computer interaction according to the situation context within ambient scenarios. 

----
STEPS

The following steps are planned:
•	Define an initial model of the A4 agent from the literature and a requirement analysis (required communicative goals and their possible multimodal expression with respect to identified ambient scenarios ; relevance of different animated agent technologies)
•	Design a wizard of Oz experiment comparing different versions of the agent (non adaptative versus adaptative) ; Tentative hypotheses to test include: the agent improves the objective evaluation (task success rate, duration, …) ; the agent improves the subjective evaluation of the user (fun, trust, reassuring) ; the dynamic management of the cooperation between modalities and the consideration of individual differences displayed by the agent improves the adaptation ; via a combination of multiple modalities that is selected dynamically (depending on user’s and objects’ locations), the agent can be used to point to real-world object ; symmetry (the user and the agent use speech and gestures) improves the interaction
•	Define an A4 model informed by the results of the WOZ (objective and subjective measures, user’s behavior) and then specify, implement and evaluate the A4 agent

----
LOCATION

This thesis will be held at CNRS-LIMSI, Orsay (www.limsi.fr), 45 mn by train from Paris.

----
FUNDING

The thesis will be funded by the ATRACO project. The salary is 1432 Euros per month.

----
CONTACT

The thesis will be co-directed by Jean-Claude MARTIN (80%) and Yacine BELLIK (20%).
For more information, please contact: Jean-Claude MARTIN 
Tel: +33 6 84 21 62 05, MARTIN@LIMSI.FR, www.limsi.fr/Individu/martin

----
REFERENCES

Dehn, D. M. and van Mulken, S. (2000). "The impact of animated interface agents: a review of empirical research." International Journal of Human-Computer Studies(52): 1-22. 

Martin, J.-C., Buisine, S., Pitel, G. and Bernsen, N. O. (2006). "Fusion of Children's Speech and 2D Gestures when Conversing with 3D Characters." Journal of Signal Processing. Special issue on "Multimodal Human-computer Interfaces". Eds.: Thierry Dutoit, Laurence Nigay and Michael Schnaider. Elsevier. 86(12): 3596-3624. http://dx.doi.org/10.1016/j.sigpro.2006.02.040

Martin, J. C. (2006). Multimodal Human-Computer Interfaces and Individual Differences. Annotation, perception, representation and generation of situated multimodal behaviors. Habilitation à diriger des recherches en Informatique. Université Paris XI. 6th december 2006. 

Ruttkay, Z. and Pelachaud, C. (2004). From Brows to Trust - Evaluating Embodied Conversational Agents, Kluwer. http://wwwhome.cs.utwente.nl/~zsofi/KluwerBook.htm

Thórisson, K. R. (2002). Natural Turn-Taking Needs No Manual: Computational Theory and Model, from Perception to Action. Multimodality in Language and Speech Systems. D. H. B. Granström, I. Karlsson (Eds.). Dordrecht, The Netherlands, Kluwer Academic Publishers: 173-207.

-----------------------------------------------------------