Enhancing Car-Following Models with Bike Dynamics for Improved Traffic Simulation

TL;DR

This paper introduces the first dedicated bicycle movement model for SUMO, improving traffic simulation accuracy by incorporating realistic bicycle dynamics based on real-world data, and demonstrating significant performance improvements over previous approximations.

Contribution

The paper presents the Realistic Bicycle Dynamics Model (RBDM), a novel model for SUMO that accurately simulates bicycle behavior using real-world data, filling a critical gap in traffic simulation tools.

Findings

RBDM outperforms existing slow-vehicle models in SUMO.

The model closely matches real-world bicycle data from SimRa.

It generalizes well across different urban scenarios.

Abstract

Road traffic simulations are crucial for establishing safe and efficient traffic environments. They are used to test various road applications before real-world implementation. SUMO is a well-known simulator for road networks and intermodal traffic, often used in conjunction with other tools to test various types of applications. Realistic simulations require accurate movement models for different road users, such as cars, bicycles, and buses. While realistic models are already implemented for most vehicle types, bicycles, which are essential for achieving safe and efficient traffic, can only be modeled as slow vehicles or fast pedestrians at present. This paper introduces the Realistic Bicycle Dynamics Model (RBDM), the first dedicated bicycle model for SUMO, addressing this significant gap. Leveraging real-world bicycle data from the SimRa dataset, the RBDM implements realistic speed, acceleration, and deceleration behaviors of bicycles in urban scenarios. The evaluation is conducted using the Monaco SUMO traffic scenario and a newly generated Berlin scenario in SUMO. The RBDM significantly outperforms the existing slow-vehicle approximation in SUMO, aligning more closely with real-world data. These results underscore the necessity of a realistic bicycle movement model for accurate simulations, given the significant differences in the movement profiles of bicycles, cars, and pedestrians. Furthermore, the model is tested for its ability to generalize to disparate scenarios and urban topologies, which is dependent on the manner and geographical region in which the SimRa data were gathered. In addition, recommendations are provided for how it could be adapted for use in different city topologies.