Generating Realistic road usage information and origin-destination data for traffic simulations: Augmenting agent-based models with network techniques

Abstract

We present a novel network approach, supported by an agent-based simulation using empirical survey results, in order to generate origin-destination data and information about the road usage of a large, urban traffic system. Additionally, we investigate congestion and its effects on road usage due to traffic jam avoidance strategies. The investigated city serves as a case study and the presented method can be easily adapted for arbitrary traffic networks. We find that the use of network techniques offers various advantages and can replace aspects that are traditionally performed by computationally more expensive methods. Our method shifts the computational efforts from individual agent interactions to more elegant network techniques, which leads to much lower computation time and better scaling properties. Results are evaluated and show high conformance with measured data, especially if congestion effects are included. Furthermore, the obtained data can be used as an input for car-following models or other types of traffic simulation to gain even more information about the investigated traffic network.