SMA
(Julien Saunier) home: english french
>research >publications >teaching >cv
Current Assistant professor at LITIS, INSA de Rouen.
Behavioural modeling and multiagent systems.
Researcher Researcher in the LEPSiS unit (Ifsttar).
Behavioural modeling and simulation applied to transport.
Postdoctoral fellowship Project SAGECE (simulation for the improvement of crisis management learning)
Modeling of the decision process of virtual autonomous agents for crisis simulation
PhD Thesis Multi-party communications and their regulation in multi-agent systems
Advisors: Suzanne Pinson and Flavien Balbo.
Supervision
  • Zaher Yamak: Detection and grouping of sockpuppets in social media. With Laurent Vercouter.
  • Lancelot Six (Ph.D., 2010-): Comparative Approach of Trucks effect on Traffic Modelling: Mathematical and Behavioural Simulations. With Zahia Guessoum and Sio-Song Ieng.
  • Kévin Darty (Ph.D., 2011-): Evaluation of the validity of the behavior of the traffic actors simulation for driving simulators. With Nicolas Sabouret.
  • Mohamed Ahmed El Iyel (Msc training, 2011): Data aggregation for the perception of the environment in multiagent systems.
Topics  
  • Decision process including physical and emotional factors
  • The effect of gender on driving behavior and on the user perception of the simulation devices
  • Simulation validity using objective and subjective data
  • MultiAgent Systems, inter-agent communications
  • The environment as First-Order Abstraction
  • Regulation and norms
Focus (thesis)  

The environment has recently emerged as a first-order Abstraction in the MultiAgent Systems (MAS) community. However, it is still mostly used for situated MASs, and/or as a support for reactive agents. One of our central theme is the unification of the concepts of perception, and direct/indirect/multi-party interaction, for cognitive agents.

In this model, the environment becomes the mediator of the MAS. Hence, a second part of our research is its impact in terms of costs and of agent knowledge. Notably, taking into account the needs of the receptors creates a need to improve the semantics associated to these interactions, in order to enable the agents to manage dynamically their interactions.