Towards a Multi-Agent System for On-Line Simulations Based on Real-World Traffic Data

Abstract

This paper presents a framework for on-line simulations and predictions of traffic states in large-scale networks, which is based on the combination of real-world traffic data with an agent-based traffic flow model. The system is applied to the urban road network of Duisburg and the freeway network of North Rhine-Westphalia. On the basis of historical traffic demand data heuristics are derived, which are combined with the current data to provide a short-term traffic forecast. Additionally, the demand for an anticipatory traffic forecast, which includes decision-making and route choice behaviour of the road users, is discussed.