Name : Ateliers de Réalité Virtuelle (Virtual Reality Workshop)

Aims : The main aim of our work is to give the entities within virtual environments the ability to behave autonomously. This raises three issues which our work endeavours to answer:

• What are the essential foundations of autonomy, and how can an entity become capable of autonomous behaviour ?
• What methods and models are used to render these environments, and the behavioural interfaces for human users, multi-sensory ?
• How can we ensure the complete immersion of human users in interaction with the autonomous artificial entities with which they must cooperate?

Our work is positioned within the context of virtual reality and multi-agent systems. Using the multi-agent systems paradigm we can consider virtual environments as being populated by autonomous entities. Augmented (or mixed) virtual reality allows humans to immerse themselves in and interact with these entities. Because of this duality, we can experiment with artificial realities in new ways; it is key to the study of complex systems as we understand them here at the team.

Responsable : Pierre DE LOOR

Participants : Matthieu AUBRY, Yvon AUTRET, Cyrille BAUDOUIN, Ronan BILLON, Cédric BUCHE, Pierre CHEVAILLIER, Nicolas COTE, Pierre DE LOOR, Frédéric DEVILLERS, Goulven GUILLOU, Frédéric JULLIARD, Vincent KOEHL, Frédéric LE CORRE, Eric MAISEL, Kristen MANACH, Nicolas MARION, Daniel MELLET, Alexis NÉDÉLEC, Mathieu PAQUIER, Mircea POPOVICI, Ronan QUERREC, Cyril SEPTSEAULT, Fabien TENCE, Jacques TISSEAU, Thanh Hai TRINH, Morgan VEYRET

Scientific Foundations :

The autonomous behaviour thematic is constructed around the development of behavioural architectures and the identification of principles concerning the autonomy of models. The identification of these principles relies on the study of the autonomy principle (as it is understood in other domains, such as the cognitive sciences) and its tranference to the virtual. The aim is therefore not to define an autonomous model, but rather the principles of a model which can become autonomous by itself. Interactive virtual learning, anticipation and the notion of co-evolution are studied on the EBV and SARA projects. The GAM work group is a forum set up for the discussion of these issues.

The multi-sensory account of virtual environments inhabited by autonomous entities relies on the design of behavioural interfaces as much from a technological point of view as a psycho-sensory one. The difficulty is to provide accounts and animations in real-time which do not affect the simulations, and which use multi-processor architecture (CPU, GPU, or even PPU). Research in this domain has been put into use in the ARéVi platform, which is also used by the EBV project.

The immersion and interaction of human users in virtual environments populated by autonomous entities, particularly artificial humanoids. Virtual humanoids are autonomous actors capable of identifying the presence and intentions of other actors, entities and users within the environment. Through numerous exchanges, these actors are able to pool information, knowledge and communiaction concepts. The goal is to achieve a coexistence of virtual actors and humans within one environment. The other thematics of the ARéVi team on autonomy, learning and anticipation, as well as the connections with research conducted at the SARA project, represent a major collaboration for this field of research.

Technical Capacities :

• Virtual Reality, augmented reality, movement synthesis and analysis, accoustics, sound perception, sound spatialisation, behavioural animation, Man-machine interaction, artificial intelligence, artificial vision, artificial learning, meta-modeling, multi-agent systems.
• Virtual reality platform, autonomoy, artificial humanoid, knowedge engineering.
• Participatory and immersive simulations, virtual training environments, intelligent tutoring systems.

Applications :

• Virtual Reality and the Arts. This application consists of artistic projects incorporating significant multi-model real-time interactions between both real and virtual artists. In this way, we can study: the efficiency of various communication methods between humans and humanoids, the behavioural realism of the actors and the evolution of the virtual actor's knowledge of his environment.
• Virtual Worlds and Virtual Communities. These applications are used with large-scale virtual environments such as historical sites, in which human actors are integrated and thus interact with the autonomous artificial entities be they human or animal, realistic or imaginary.
• Virtual Learning Environments. These applications stem from the Mascaret platform. In Mascaret, the training environment and the learners' and instructors' activities are described as the instances of a meta-model, which facilitates their definition and means that these virtaul learning environments can have adaptive tutoring systems. This work is conducted in affiliation with cognitive psychologists and educational science researchers from the SARA project (Virtual Learning Environment work group). They are also based on the development of evaluation methods in the activities of human subjects, a domain which has so far not been examined in depth in terms of virtual reality. One application of Mascaret is the virtual work surface project.

Collaborations :

• University of Southern Brittany, Valoria : movement synthesis and analysis.
• University of Technology of Compiègne, HEUDIASYC : virtual reality and knowledge
• INRIA - Rennes, Bunraku : virtual training environment.
• University of Constanta, Romania, Cerva team: reconscruction of historical sites (TOMIS project) and virtual training environments (the EVE project). This collaboration leads to frequent research exchanges.
• LISyC is a member of AF-RV, the French association for virtual, mixed and augmented reality and 3D interaction.

Realization :

• Platforms for the the interactive simulation of autonomous entities
  • ARéVi : virtual reality platform.
  • oRis : programming and agent-oriented simulation environment.
• Libraries
  • HLib 2: animation of artificial humanoids for ARéVi.
  • ArFCM : fuzzy cognitive maps for ARéVi.
• Applications of research outcomes
  • ARéVi-Road : virtual training environment for driving in traffic. ANR project led by INSERR, LISyC and CervVAL.
  • behavioRis :reactive behaviour simulation. European projects: Premecs II and Necessity.
  • Mascaret : Virtual environments for learning.
  • Gaspar : Masacret application for the simulation of air traffic on aircraft carriers. Project conducted for the DCNS.
  • GVT : Generic Virtual Training : virtual training environment for maintenance. This project was conducted in collaboration with the Irisa (Siames team) for Nexter Systems.
  • Virtual work surface (Paillasse virtuelle) : a virtual training environment dedicated to practical experimentations in engineering and experimental science.
  • Sirene : virtual guide in augmented reality.
  • CoPeFoot : Collective Perception in Football.
  • Enactive Models