Groupe Architectures et Modèles pour l'Interaction

Presentation du groupe AMI

The Research team « Architectures and Models for Interaction » (AMI), iinitiated in 2000 at LIMSI-CNRS, is dedicated to the study of the interactional phenomena that occur between humans and computer systems. The Scientific works of the AMI team are organized along four main topics. The structuration of AMI research was defined in 2011 to capture what we consider as being an important evolution in the domain of human/machine interactions. It is composed of four topics

• Topic « Image and interaction » deals with image processing for augmented reality, medical applications and robot vision. It is an area where the relationship between data processing and acting in the world is particularly important.
• Topic « Ambient and interaction » deals with new interaction issues within the context of ambient environments (interaction adaptation, merging of virtual and physical worlds, ambient modalities) and with conversational and social issues. It is where we think we can gain empirical evidence on how virtual and physical objects can be combined in order to produce meaningful, emotionally charged experiences for helping people understand the world in which they find themselves.
• Topic « Haptic interaction and communication » deals with the characterization of abstract environments, collaborative and emotional haptics and results in application to learning. It also provides us with a testbed for questions concerning the construction of mutual trust and confidence when learning to do new things together.
• Topic « Interaction with tactile surfaces » deals with the support of collaborative activities, the study of the potential of multi-touch inputs and results in several applicative fields of tactile surfaces. It also allows us to address the question of the extent to which tangible user interfaces help people understand what is required of them in a micro-world situation.

The Research team « Architectures and Models for Interaction » (AMI), started working at LIMSI in the year 2000 on understanding and experimenting new forms of human-machine interactions. Initially focused essentially on Graphical User Interfaces (GUI), the emergence in recent years of new classes of users, new software and hardware tools and new interaction environments has shifted our attention from interfaces to people of the general public who are engaged with others in highly dynamic physical and virtual environments. Indeed,  users of current computer systems are now mainly “ordinary people”. They generally have only a limited knowledge of how their computers work, but they increasingly use their technical devices to engage in socio-ludic activities; expect that these devices will learn to recognize and anticipate upon their on-going and constantly changing needs; and are happy when information is embodied in digital objects which convey attitudes and emotions that help to make that information meaningful.

We are consequently working to replace the “key-mouse-window” paradigm with an “in the world” paradigm. Acting in the world often means dealing with uncertainty by representing complex phenomena in new ways. However, understanding the world doesn’t necessarily mean just processing data. Often an emotional and/or aesthetic experience is just as important, as is working hands-on with tangible user interfaces to manipulate physical objects, surfaces or spaces. Changing the world is a third theme: it requires working with others but how are solid relationships and mutual trust built up and maintained over time? Finally, the fact that more and more computing power is being incorporated into everyday objects which are capable of communicating with one another and with the people that use them is changing the world as we know it. Virtual agents are acting on their own, physical phenomena and informational phenomena are being interlaced in ambient environments. But what is the impact, what are the dangers and what are the ethical questions that have to be raised when working to build architectures and models for supporting real-world interactions?

AMI is composed of people trained in image processing, multi-modal data processing, agent technologies, signal processing, sociology and AMI cooperates widely as well, both inside LIMSI with other teams working notably on robot vision and natural language processing and externally, with industrial and scientific partners in Digitéo and other national and international research frameworks. We are in a fast moving and highly competitive field which explains why we tend to update our multidisciplinary priorities every two or three years. What follows is the structure of our research program as it was defined in the fall of 2011.