Simulation of Traffic Network Re-organization Operations

摘要

In this paper we present a model for an Agent-based intelligent Transportation System (ATS). In ATS, the traffic environment is partitioned into areas controlled by specialized agents called service managers. These agents are supervised by a traffic manager agent whose responsibility is to identifying global traffic management strategies. Intersections and vehicles are also equipped with agents. ATS is based on the premises that traffic management is the result of interactions between the various types of agents. We give an overview of MATISSE 2.0, a large-scale multi-agent based traffic simulation platform developed to test traffic scenarios in ATS. We discuss the underlying structure of the traffic network in MATISSE and provide a detailed discussion on the atomic network re-organization operations, i.e., incoming traffic flow elimination, road reversal and crossing elimination. Unlike many existing simulators which execute pre-defined network strategies, MATISSE allows ATS agents to select and execute network re-organization operations dynamically at run-time. The simulation of an evacuation scenario in a traffic network consisting of 384 roads and 64 intersections shows that the traffic re-organization operations improve evacuation time but are not be optimal.