Cycling is increasingly promoted as a mode of active mobility, especially in urban areas, because it supports climate, health, and liveability goals. In Austria, this promotion has been successful, with cycling increasing significantly in recent years. However, cyclist safety has not improved at the same pace. While most other modes of transport have become safer, the number of accidents involving cyclists has increased by 81 percent over the last twelve years. This suggests that factors contributing to cyclist accidents are still not sufficiently understood, especially regarding street infrastructure and its design.
This thesis examines this problem through hazardous road locations in Vienna. It asks how street characteristics at these locations influence cyclist safety and what this implies for the further development of Austrian cycling guidelines. A sequential mixed-methods design was used, combining Getis-Ord Gi* hot spot analysis of police-reported cyclist accident data from 2022 to 2024 with a qualitative analysis of selected locations. The quantitative analysis identified 91 hazardous road locations, of which 13 were selected as representative cases for different accident groups and assessed against the current RVS Bicycle Traffic.
The results show that many hazardous road locations have recurring accident patterns. These patterns were often connected to intersections, street layout and space allocation, public transport-related conflict points, and visibility conditions. The thesis shows that the RVS already provides a strong basis for safe cycling infrastructure, but that some criteria need to become clearer, more measurable, and more binding to better address recurring cyclist accident patterns.
... Read more

Automated driving systems increasingly rely on learning-based components, which makes their safety evaluation challenging. Runtime monitoring provides a complementary safety layer by flagging anomalies before they escalate. However, current monitoring approaches are predominantly rule-based, which face limitations in complex conditions. Recent advances in vision/large language models, with their capacity for reasoning and context-aware scene interpretation, offer a promising alternative. Nevertheless, existing research primarily focuses on monitoring perception and largely neglects the decision-making and planning (DMP) layer, despite its critical importance. There is also a lack of an integrated framework that unifies monitor outputs into actionable severity levels. This paper addresses these gaps with two contributions. First, we propose a conceptual monitoring framework with a minimally intrusive side-car architecture, layered organization, and a unified four-failure-mode semantics. Second, we instantiate this framework for trajectory prediction within the DMP layer and conduct a controlled comparison of three monitoring paradigms: (i) a non-AI statistical baseline, (ii) a fine-tuned CLIP-based vision-language classifier, and (iii) a zero-shot multimodal LLM. Evaluation results on nuScenes dataset reveal that statistical monitoring struggles to detect safety-critical failures, while the vision-language monitor excels at low-to-mid severity detection, and the LLM-based monitor achieves the strongest performance on the most severe failure modes.
https://github.com/HaodongLi-Hub/How_To_Monitor_AV_Using_AI
... Read more

This study investigates the factors influencing passenger distribution on railway platforms in the Netherlands, using Eindhoven Central Station as a case study. The objective is to quantify how spatial characteristics, environmental conditions, and human behavioral factors jointly shape passenger distribution and to derive actionable recommendations for platform design and management.
A combination of descriptive statistics, spatial analysis, and discrete choice modeling was applied to a high-resolution dataset comprising 142,256 sensor-based observations collected under 49 situational scenarios. The platform was discretized into spatial cells characterized by distance to entrances, information boards, and the track edge, proximity to seating, leaning areas and kiosks, passenger density, lighting conditions, and weather.
Descriptive analyses reveal systematic clustering near entrances and comfort-related facilities, confirming the central role of accessibility and physical support in waiting behavior. Passengers generally avoid track-adjacent areas, while moderate social clustering occurs at intermediate densities. Environmental conditions further influence spatial patterns, with adverse weather and poor lighting reinforcing concentration in sheltered zones.
A multinomial logit model identifies nine statistically significant determinants of waiting location choice. Seating exhibits the strongest positive effect, followed by entrance proximity and leaning facilities, highlighting comfort and accessibility as primary drivers. Safety considerations reduce the attractiveness of areas near the track, although this effect weakens under favorable lighting and weather conditions. Areas near kiosks and information boards are avoided, indicating the disutility associated with congestion and circulation conflicts. The model demonstrates strong predictive performance and reproduces observed passenger distributions with high accuracy.
The findings show that platform waiting behavior reflects structured trade-offs between comfort, safety, accessibility, congestion avoidance, social context, and environmental conditions. Based on these results, design recommendations are proposed, including redistributing seating and leaning facilities, dispersing entrance flows, relocating kiosks and information boards toward circulation corridors, applying adaptive lighting strategies, and maintaining clear safety buffers near the track edge. The study provides a behavioral and empirical foundation for improving comfort, safety, and operational efficiency at Dutch railway stations. ... Read more

The rapid increase of e-bike use in the Netherlands has intensified speed heterogeneity on cycling infrastructure. Although this trend contributes positively to sustainable mobility, it may negatively influence cyclists’ perceived safety, which in turn affects their route choices and willingness to cycle. Existing literature recognizes the role of infrastructural, environmental, traffic-related, and individual factors in shaping perceived safety, yet the specific influence of speed differences in bicycle-only environments remains insufficiently studied. This research investigates the extent to which speed heterogeneity affects perceived cycling safety and how this effect interacts with other contextual attributes. A video-based stated preference experiment was designed in which respondents evaluated overtaking events filmed in real urban locations, manipulated across lane design, lane width, cycling density, and speed differences. A panel Mixed Logit model was estimated, supplemented with scenario ranking tasks. Results show that greater speed differences consistently reduce perceived safety, especially under congested conditions. Two-way lanes are perceived as less safe than one-way lanes, though wider lanes mitigate this effect and experienced cyclists express higher comfort levels. Findings underline the importance of managing speed heterogeneity through infrastructural design and policy interventions, including multi-paced cycling lanes, speed awareness campaigns, and targeted safety reporting mechanisms. The study demonstrates the value of video-based stated preference methods for analyzing subjective safety and contributes to policy debates on designing cycling infrastructure for increasingly diverse urban cyclists. ... Read more

Walking is important for elderly people as it allows for community interaction, access to resources, and helps maintain mobility and good health. Previous research found that barriers in public space and individual factors, such as reduced mobility, influence safety perceptions of elderly when walking and can eventually affect their willingness of walking. While the municipality of Amsterdam monitors the safety perception of residents, few insights are provided on how elderly perceive their safety while walking and how different determinants are of influence. This research explores individual safety perceptions of elderly and their influence on walking by combining a policy document analysis, semi-structured interviews and Q-methodology. The interviews and Q-methodology were conducted with 14 elderly residents living in three city districts in Amsterdam. Two shared perspectives on safety perception were found, with one emphasizing the importance of respectful behavior from traffic users and the other highlighting good sidewalks that are not slippery. Next to this, sidewalk barriers, traffic speed, and public lighting were important aspects. One interesting finding was the strong negative influence of the behavior of cyclists and micromobility users on the safety perceptions of participants. The policy analysis showed different municipal approaches in relation to walking and revealed that limited space for redesign, limited maintenance budgets and a lack of clear policy on micromobility use currently form the greatest barriers to enhancing safe walking among elderly. Findings of this study help to understand the safety perceptions of elderly residents in more detail and can help inform future municipal policy aimed at encouraging walking among this group. ... Read more

The increasing prevalence of micromobility vehicles in urban environments has raised concerns about safety in shared cycling spaces. This study examines the overtaking behavior of e-scooter and e-bike riders to inform traffic management strategies and infrastructure development. A controlled experiment was conducted using strategically placed cameras to track vehicle trajectories and inertial measurement units (IMUs) to capture roll data. Extensive data processing ensured accuracy and synchronization of trajectory and IMU information. Key findings reveal that e-bikes overtaking e-scooters initiate maneuvers from greater distances but maintain smaller lateral distances compared to e-bikes overtaking e-bikes.
Lateral position differences showed a stronger correlation with speed difference than longitudinal position differences. The highest roll rates and angles occurred during the overtaking phase. Pre-overtaking, higher roll rates and angles were observed when e-bikes overtook other e-bikes, indicating greater control adjustments. No significant gender differences were found in overtaking behavior. However, in non-interactive scenarios, male e-scooter riders traveled at higher speeds than females, while no gender differences were observed among e-bike riders. These results provide insights into the complex interactions between different types of micromobility vehicles during overtaking maneuvers. The findings underscore the need for targeted safety interventions and infrastructure improvements to mitigate risks associated with shared cycling spaces, ensuring safer coexistence of micromobility users and conventional cyclists in urban environments. ... Read more

This study explores data-driven bicycle policy development using “Talking Bikes” data from RingRing and Tracefy. These datasets contain high-resolution GPS-based trajectory data of cyclists, including position, speed, direction, and trip information. The goal of the study is to improve data quality and to explore potential policy applications of large-scale bicycle mobility data.

Data quality improvements include map-matching, interpolation and extrapolation of trajectories, and outlier detection. In addition, multiple use cases are developed to demonstrate how processed cycling data can support urban mobility policy. These include real-time traffic monitoring, area-based network utilisation (GGB+), transport demand estimation, network design, and policy evaluation.

From these applications, key performance indicators (KPIs) are derived across categories such as accessibility, safety, reliability, health, environment, and equity. Examples include flow, speed, travel time, route choice, stops, incident risk, and exposure. These indicators enable detailed analysis of cycling behaviour in different temporal and spatial contexts, such as peak versus off-peak hours and weekday versus weekend patterns.

The study demonstrates that high-resolution bicycle data can support a wide range of policy applications, from real-time traffic management to long-term infrastructure planning. By improving data quality and systematically structuring mobility indicators, Talking Bikes data can provide valuable insights for more effective and evidence-based bicycle policy development. ... Read more

In crowded pedestrian environments, individuals often resort to rotating their bodies as a strategy to avoid collisions. Surprisingly, this rotational behavior, despite its significant implications for crowd capacity, has received relatively little attention in research. This study seeks to fill this gap by diving into the complicated world of pedestrian rotation behavior, particularly in the context of high-density bidirectional and crossing flows.
Drawing upon data gathered from the CrowdLimits experiments, we start the exploration of how various factors impact the rotation behavior of pedestrians. Our investigation covers crowd density, the fundamental movement scenarios (bidirectional and crossing flows), flow ratio, and the influence of disturbances within the crowd under different scenarios.
Our key findings reveal that all these factors play a role in shaping the frequency of rotations within a crowd. However, the extent and precise conditions under which these factors influence this subject demand further in-depth research and exploration.
In essence, this study addresses the fundamental question: How does shoulder rotation behavior vary concerning macroscopic crowd characteristics, including crowd density, flow ratio, and movement patterns like bidirectional and crossing flows? Through this research, we hope to highlight the complex interplay between these factors and the rotational strategies operated by pedestrians, ultimately enhancing our understanding of crowd dynamics. ... Read more

Bicycles have an important role to play in the transition towards a more sustainable mobility. In order to achieve the modal shift towards bicycles, more must be done to accommodate cyclists. Although controlled intersections do increase the (perceived) safety of crossings with motorized vehicles, they are seen as major obstacles, and cyclists tend to avoid them when possible. The negative effects of controlled intersections for cyclists may be reduced by new methods of intersection control. This thesis combines the concepts of the connected environment and structure free control, to design an intersection controller that uses a genetic algorithmto determine the optimal signal plan, hereafter referred to as the SFGA controller. The controller is designed for an isolated intersection and considers car drivers and cyclists. Desires of cyclist with regard to controlled intersections, are identified by means of a literature review on the determinants of bicycle use, which are then projected on the controlled intersection. A traffic system model, based on validated models found in literature, is set up and a design for the structure free controller is proposed. A set of control objectives is proposed, including different metrics related to the desires of cyclists and car drivers. Objective function weights can be varied to achieve different levels of cyclist prioritization. A maximumwaiting time of 100 seconds is enforced in order to prevent prioritization of cyclists to result in unreasonable delays for car drivers, because red light running probabilities increase at larger waiting times. The performance of the structure free controller is evaluated for 15, 35 and 45% traffic saturation (percentage of intersection capacity), by means of a simulation based case study. The designed controller is benchmarked to vehicle actuated control (VA). VA has a cyclic, fixed control structure in which green times of movements are flexible and depend on the queue size. SFGA is benchmarkedwith an equalweight for the delay of car drivers and cyclists, and no weight included for the number of stops. The effect of incorporating weights that prioritize the desires of cyclists over those of car drivers is investigated. The SFGA controller results in average delays 1.8, 2.7 and 3.0 times lower than VAC for each of the evaluated traffic saturation levels. The number of stops is 1.9, 2.3 and 3.1 times lower. Including weights in the objective function to explicitly prioritize cyclists, results in even lower average delays and number of stops for cyclists. As is to be expected, this comes at the cost of additional delays for car drivers, especially for higher traffic saturation. The better performance of SFGA is attributed to two main differences between the controllers. First of all, the structure free aspect allows for a larger degree of freedom to choose more effective combinations of traffic lights to show green at the same time, instead of following the fixed sequence of VA. Additionally, the controller allows traffic that otherwise would experience the largest total delay to cross first, even if this means delaying some travellers in close proximity of the traffic light. This contrary to VA, that extends green time based on detected traffic in the active block. Without inclusion of weights that prioritize the desires of cyclist over cars, the controller already tends towards prioritization of the cyclists. This is caused by the controller considering the number of travellers that are influenced by its’ control decisions, combined with the higher traffic densities, that can be expected on bicycle paths in urban areas. Weights to prioritize cyclists can be included to include more priority, for example when bicycle traffic volumes are low. This work implicates that, in order to better serve the cyclists, it is not explicitly required to prioritize cyclists over cars. In areas with large volumes of cyclists, considering the number of travellers and their proximity to the traffic light can already result in cyclists being served better. This work could be used as a starting point or inspiration to design and eventually implement more cyclist oriented intersection controllers. Improvements for the controller and extensions for the research scope are proposed that are required for the controller to be suitable for practical implementation in the real world. If a future version of the controller is to be implemented, it will reduce the negative effects of controlled intersections on cyclists, thereby making the bicycle a more suitable replacement for the car. ... Read more

In 2019 a virus surfaced in China that would later cause a pandemic: the coronavirus. Across the world measures were taken to reduce the spread of the virus. Public Transport (PT) had a unique position, being both an essential service and a place where the virus can easily be spread. This research investigates the measures taken in PT in the Netherlands and internationally in an objective manner. No assessment is made of any kind of any measure. In the Netherlands, the Outbreak Management Team collaborated with the public transport association OV-NL to ensure every companies response was similar. Through an interview with the chairman of OV-NL, Pedro Peters, more information was gathered on this particular aspect of the research. Besides taking measures to reduce the virus, mainly by reducing or avoiding contact (both between people and with surfaces), companies reduced their service levels. This reduction was a response to the large decrease in passenger demand and partly to more of the personnel reporting sick. This research investigated this reduction of service throughout the first period of the virus outbreak in the Netherlands (27th of February – 1st of June). Besides the Netherlands, measures taken in other countries were also investigated, showing many similarities. However, other measures were taken as well, like checking the temperature of passengers in for example Thailand, China or Iran or using a proximity app which happened in Singapore. Not every country experienced the virus outbreak at the same time. In the research, data by the Oxford University and Blavatnik School of government was used to create world maps that visualize the closing of public transport for all days of the outbreak up to June 1st. The same was done to create a map for lockdowns and using data found through an online search a world map visualisation was made about the mandatory use of face masks in public transport. The maps show a general trend: following the virus, but it is clear that some countries responded early or late and that the correct response to the virus was unclear. Especially the face mask visualisation shows a more random pattern. So while within the Netherlands the response to the virus was uniform, internationally the response was quite random. This research can be used to further research the measures taken and assess the measures with regards to for example effectiveness, safety or comfort. ... Read more

To intrigue more people to cycle, it is essential to assess the current design of the cycling infrastructure to guarantee the efficiency and safety for the users. A credible simulation of cyclists is helpful to evaluate different designs of infrastructure and point out the problem. A valid simulation model requires proper verification and validation. However, there was hitherto no standard procedure to conduct model verification, and limited studies were found related to the verification of cyclist simulation models. To fill this gap, this study develops a verification framework for traffic simulation models based on the methodologies used in the literature. It also briefly discusses the difference to verify a cyclist simulation model. The framework is assessed by applying to a case study of the cyclist-queue-formation simulation model. The results prove the applicability of the framework to verify a cyclist simulation model. The organised structure of the framework provides a systematic measure for the verification of traffic simulation models. Furthermore, the cyclist-queue-formation simulation model is verified within the study area, where its conceptual model is proposed. The room for improvement of the simulation model regarding the relatively slow correction behaviour in steering and pedalling of the cyclists was also indicated. ... Read more