Abstract EN:

For several years, distributed virtual environments (DVEs) have been increasingly popular in the form of online video games, training simulators or collaborative virtual environments. Nowadays, such applications often include computer simulated autonomous entities to provide densely populated virtual worlds. However, most current DVEs involve static behavioral models which are not much sensitive to bandwidth limitations, packet loss, latency or jitter. In this context, the first achievements presented deal with the simulation of autonomous adaptive entities in a High Level Architecture-based DVE. They raise some issues related to the effects of distribution on such entities' capabilities to make complex behaviors emerge. To minimize those drawbacks, an approach based on management areas is put forward to reduce the amount of interactions that rely on network communications. This document finally describes a mechanism designed to optimise the dead reckoning algorithm which goal is to reduce the amount of exchanged messages using entity state prediction.