Stéphane Donikian

Head of the Bunraku Project-Team 

Campus de Beaulieu
35042 Rennes Cedex, France
Tel: +33 2 99 84 72 57
Fax: +33 2 99 84 25 27

My research has been performed during those years at IRISA (Information Technology and Non-deterministic Systems Research Institute), a publicly funded research laboratory. IRISA is organized in around thirty research projects or teams. I am currently leading the Bunraku team, which is a new research team that inherits the long experience made by the SIAMES team conducted by Bruno Arnaldi over the preceding twelve years. Bunraku is an INRIA research project-team currently composed of 45 persons from five institutions (INRIA, CNRS, University of Rennes 1, INSA Rennes, ENS Cachan antenne de Bretagne). The synthetic definition of the research area addressed by Bunraku is: Perception, decision and action of real and virtual humans in virtual environments and impact on real environments.

Our challenges are to:

To do so, we are working on three complementary research themes:

To be able to gather all those works inside applications, we have developed a VR middleware which is available in its third version under an Open Source License (see

My own research activity concerns mainly the second and third research themes. Concerning Expressive Autonomous Characters, I am conducting some research activities concerning the modelling and simulation of individual and collective human behaviours. The principal key words of the whole work done so far on virtual humans are: how to model it, animate it, control it, and define its interactions with its environment and its congeneric. Historically, the first application domain concern multimodal traffic simulation in urban environments. More recently we have enlarged the application spectrum to industrial products and process, video games, narrative and interactive virtual worlds, and virtual training. Several research projects have been conducted with industrial and academic partners, as described below.

Behavioural models offer the ability to simulate autonomous agents like organisms and living beings. Psychological studies have showed that the human behaviour can be described by a perception-decision-action loop, in which the decisional process should integrate several programming paradigms such as real-time, concurrency, and hierarchy. Building such systems for interactive simulation requires the design of a reactive system treating flows of data to and from the environment, and involving task control and preemption. Since a complete mental model based on vision and image processing cannot be constructed in real time using purely geometrical information, higher levels of information are needed in a model of the virtual environment. For example, the autonomous actors of a virtual world would exploit the knowledge of the environment topology to navigate through it. Thanks to this knowledge, autonomous virtual actors can behave like pedestrians or car drivers in a complex city environment.

You can find here my list of publications and here different documents (papers, images and videos) related to my research activities.