Modeling cyclists' safety and mobility using crowdsourced and simulator data

Cycling plays a crucial role in sustainable urban mobility, and many cities have increasingly recognized and prioritized cyclists’ safety and mobility in urban planning. Furthermore, advancements in emerging technologies have facilitated access to larger and more detailed datasets on cyclist behavior and safety, enabling a more comprehensive analysis of safety and mobility patterns. Despite these developments, certain aspects of cycling safety and mobility remain underexplored. Uncovering new insights in these areas has the potential to significantly inform decision-makers in designing more effective bicycle-friendly urban policies and interventions.

This PhD thesis aims to examine cyclists’ safety and mobility at both the street and route levels by utilizing large-scale crowd-sourced data, cycling simulator experiments, and supplementary datasets. Part I (Paper 1 and 2) investigates, at the street level, the infrastructure characteristics associated with bicycle near-crashes and crashes, as well as the impact of near-crash-prone infrastructure on cyclists’ mobility and perceived safety whereas Part II (Paper 3 and 4) at route level explores how safety considerations influence cyclists’ route choices and how infrastructure design affects cyclists’ accessibility.

Part I focuses on the impact of infrastructure on cyclist safety and mobility at the street level. The first paper uses quantitative modelling to investigate the relationship between different infrastructure types and cyclists’ risk of near-crashes, using large-scale crowdsourced accelerometer data, as well as crashes recorded in police reports. The findings indicate that near-crashes occur more frequently in pedestrian-oriented shared spaces but do not necessarily result in severe crashes. In contrast, intersections and roundabouts pose significant risks for both near-crashes and crashes.

To gain a deeper understanding of cyclists’ behavior and perceived safety in near-crashprone infrastructure, the second paper examines cyclists’ mobility and subjective safety at shared paths and unsignalized intersections under various infrastructure designs and traffic flow conditions using cycling simulator experiments. The findings indicate that cycling speed and perceived safety are significantly influenced by path width, vehicle speeds, and the flow of pedestrians and bicycles. These insights highlight the critical role of optimizing infrastructure design to improve both safety and mobility in urban shared spaces.

Part II examines cyclists’ mobility, including route choices and accessibility, at the network level. This section also incorporates safety-related factors and insights identified in Part I.

The third paper explores cyclists’ route choice behavior incorporating safety factors using crowd-sourced large-scale trajectory data with various modeling approaches. The safety factors includes expected number and rate of bicycle near-crash and crashes, which were obtained from The first paper. The findings indicate that cyclists tend to avoid route segments with a high rate and expected number of both near-crashes and crashes, but the impact of expected near-crashes on route choice behavior is less pronounced compared to that of expected crashes. The results also indicate significant inter-respondent heterogeneity.

The fourth paper Investigates cyclists’ accessibility to their destinations by employing detour ratios as indicators, using large-scale crowd-sourced trajectory data combined with spatial modeling approaches. This study measures realized detour ratios (RDRs), reflecting network constraints and intentional behavior, and behavioral detour ratios (BDRs), capturing only intentional detours. The analysis conducted in Copenhagen reveals lower RDRs along well-connected radial corridors equipped with dedicated bicycle infrastructure, while the city center and recreation areas exhibits slightly higher BDRs.

This thesis makes a substantial contribution to cycling safety and mobility research, with results that could lead to important policy implications. By means of near-crash analysis, simulation-based infrastructure testing, and aggregated analysis of route choice and accessibility, this research provides valuable insights for decision-makers. Enhancing cycling safety and mobility through providing dedicated bicycle infrastructure, optimizing shared space designs, and carrying out data-driven assessments can create more cyclist-friendly cities. Ultimately, by advancing knowledge on cycling safety and mobility trade-offs, this research supports the development of urban environments where infrastructure balances
safety and mobility, fostering more sustainable transport systems. Future research should integrate diverse data sources, refine safety-mobility trade-off models, and evaluate the long-term effects of infrastructure investments across different urban contexts.