Cognitive and affective architecture for social Human-Robot Interactions

Thursday
1
Sep
2016
Offer type: 
Duration: 
4 months
Workplace: 
Host team: 
Contact person: 
Offer details: 
Research Engineer / post-doc position: 
Cognitive and affective architecture for social Human-Robot Interactions

Research context:
Most robots in use today have an industrial or military use. However service robots for daily life now challenge this professional usage. In 2012, about 3 millions service robots for personal and domestic usage were sold. These sales mainly include vacuum robots and lawn mowing robots. However, a new type of robots is emerging, aiming at assisting people during their daily life activities. One can think of assistive robots taking care of elderly people, or pedagogical robots for children. More generally these robots are called “companion robots” because their main mission is to support and assist people in their everyday life activities and to keep them company. One of the specificities of such robots is that they interact more and more closely with their human users, and their value is much more on social than physical interaction [1].  By close, we mean that robots must share not only the same physical space but also goals and beliefs to achieve a common task through their interactions.

Research challenges:
During the past decades, research in robotics has mainly focused on fundamental skills such as robust perception, navigation and catching or moving things. One of the challenges is now to endow our companion robots with subtle and smart abilities such as understanding and reasoning, emotion detection and expression, empathetic behaviour,... They should also interact intuitively and easily through speech, gestures, and facial expressions.

In spite of the numerous contributions in the field of cognitive architectures, (see good reviews for example [11],  [12]), most of them are generic and few can really deal with the complexity of human-robot interactions (HRI). They are not tailored to meet the specific needs of social HRI, such as handling emotions, language, social norms...

That is why developing a cognitive architecture for social robots able to take into account the complexity of interactions with humans still remains a real challenge. Such an architecture requires various features: emotions, non-verbal aspects of interaction, reactive and deliberative levels (fast emotional answer versus slower and more deliberate answer), explicit manipulation of mental states (enabling self-explanation)...

Objectives:
This research follows our previous work in three research projects funded by the French national research Agency (ANR): ANR CECIL project, ANR MOCA project and ANR SOMBRERO project. These projects have contributed to the ongoing development of our Cognitive and Affective Interaction-Oriented architecture called CAIO (see Best Late Breaking Report Nomination paper [10]).
The goals of this research are :
· to fully implement this architecture. A first version of an implementation already exists but needs some improvements.
· to design and to implement a relevant scenario, using a lightweight humanoid robot (for instance a Nao robot).

 
University, laboratory, team
· University of Grenoble-Alps: http://www.univ-grenoble-alpes.fr/en/ 
· Grenoble Informatics Laboratory : https://www.liglab.fr
· MAGMA team : http://magma.imag.fr/

Supervisors
· Sylvie Pesty (sylvie.pesty@imag.fr) Grenoble Informatics de Grenoble http://magma.imag.fr/content/sylvie-pesty

 
Co-Supervisors
· Carole Adam (carole.adam@imag.fr) Laboratoire d’Informatique de Grenoble http://magma.imag.fr/content/carole-adam       
· Damien Pellier (damien.pellier@imag.fr) Laboratoire d’Informatique de Grenoble http://magma.imag.fr/content/damien-pellier   

Bibliography
 
[1] Johnson, D.O., et al., Socially assistive robots: a comprehensive approach to extending independent living. International Journal of Social Robotics, 2013: p. 195-211.

[2] Cabibihan, J.-J., et al., Why Robots? A survey on the roles and benefits of social robots in the therapy of children with autism. International Journal of Social Robotics, 2013. 5(4): p. 593-618.

[3] Fasola, J. and M. Mataric, A socially assistive robot exercise coach for the elderly. Journal of Human-Robot Interaction, 2013. 2(2): p. 3-32.

[4] Mazzoleni, S., et al., Acceptability of robotic technology in neuro-rehabilitation:    Preliminary results on chronic stroke patients. Computer Methods and Programs in Biomedicine, 2014: p. 1-7.

[5] Tencé, F., et al., Stable growing neural gas: A topology learning algorithm based on player tracking in video games. Appl. Soft Comput., 2013. 13(10): p. 4174-4184.

[6] Rivière, D., C. Adam, and S. Pesty. A reasoning module to select ECA's communicative intention. in International Conference on Intelligent Virtual Agents (IVA). 2012. Santa Cruz, CA. p. 447-454.

[7] Scherer, K.R. and H. Ellgring, Are facial expressions of emotion produced by categorical affect programs or dynamically driven by appraisal? Emotion, 2007. 7: p. 113-130.

[8] Dulac, A., et al. Learning useful macro-actions for planning with N-Grams. in IEEE International Conference on Tools with Artificial Intelligence (ICTAI). 2013. Washington, DC.
 
 [9] G. Milliez, M. Warnier, A. Clodic, R. Alami. A framework for endowing an interactive robot with reasoning capabilities about perspective-taking and belief management. The 23rd IEEE International Symposium on Robot and Human Interactive Communication, 2014 RO-MAN, 2014.
 
[10] W. Johal, D. Pellier, C. Adam, H. Fiorino, and S. Pesty. A Cognitive and Affective Architecture for Social Human-Robot Interaction. In Proceedings of the Tenth Annual ACM/IEEE International Conference on Human-Robot Interaction, Extended Abstracts (HRI'15 Extended Abstracts). ACM, New York, NY, USA, 71-72, 2015.

[11] H.-Q. Chong, A.-H. Tan, and G.-W. Ng. Integrated cognitive architectures : a survey. In: Artificial Intelligence Review, vol. 28 (2), pp. 103–130, Feb. 2009. 

[12] Thòrisson, Kristinn and Helgasson, Helgi. Cognitive Architectures and Autonomy: A Comparative Review. In : Journal of Artificial General Intelligence, vol 3, pp. 1—30, 2012.