Proceedings (in PDF)

Invited speakers

Machine-learning - Decision making

Intelligent and constraint-based scene modelling

Information retrieval and visualization

Declarative Modelling

Genetic algorithms

Hardware

Short papers

Classifier system based autonomous vehicles in HLA distributed driving simulations

Samir TORKI, Patrice TORGUET, Cédric SANZA, Jean-Pierrre JESSEL

Virtual Reality & Computer Graphics Research Group, IRIT, Toulouse (France)

Abstract:

Existing driving simulations generally involve autonomous agents based on finite state machines or on expert systems. Such models require designers to describe exhaustively how agents must behave and lead to very long and demanding developments. They also require to be completely redefined when new kinds of entities have to be integrated to the simulation. Simulating autonomous entities also generally deals with the problem of involving enough entities to recreate dense traffic.
At first, this paper presents how we created autonomous agents based on classifier systems which enable to define their behaviour through goals (“what to do”) instead of transitions or rules (“how to do it”). Then, it shows how we created a distributed multi-user simulation using the High Level Architecture and how it enabled us to distribute easily the management of autonomous agents.

Key-words:

High Level Architecture, Classifier systems, driving simulation, distributed simulation.

Paper:

Paper08.pdf