Active mobility modes such as walking and cycling are increasingly emphasised in urban transportation due to their sustainability and health benefits. However, the growing use of these modes heightens the potential for conflicts, especially where pedestrians and cyclists share limited space. Understanding the interaction between these users is essential for designing safe and efficient infrastructure. This study investigates pedestrian behavioural responses when crossing bike paths with oncoming cyclists, with the goal of enhancing the accuracy of pedestrian modelling in the MassMotion simulation tool developed by Arup.
Focussing on sideways crossing scenarios, which are critical due to their frequency and potential danger, the research addresses how pedestrians adapt their movement patterns in response to cyclists' proximity and trajectory. Unlike previous studies that primarily examine head-on or rear-end conflicts, this research utilises post-encroachment time (PET) as a conflict measure to capture a broader spectrum of interactions. Trajectory data from smart sensors at two intersections on the TU Delft campus were used, encompassing over 289 000 trajectories collected over a month.
The study isolates 7,310 pedestrian interactions with cyclists, with particular attention to 4,780 one-on-one crossing events. It distinguishes cases where pedestrians crossed either before or after a cyclist and compares them to non-crossing trajectories. Results indicate that pedestrians are more likely to stop when the PET is between 0–3 seconds, particularly when crossing after a cyclist, with stopping occurring in about 35% of such cases. The average stopping distance was found to be just over 3 meters, and slightly less (2.5 meters) when pedestrians walked alongside the bike path before crossing.
Additionally, the study introduces a method for predicting PET values in real-time, allowing for a dynamic understanding of pedestrian decision-making. When predicted PET values fall below 1 second, indicating a potential collision, pedestrians are more likely to yield by stopping or slowing down. Deviation analysis further shows that around 20% of pedestrians crossing behind cyclists veer off their straight path, often moving slightly toward the cyclist to complete the crossing sooner.
These behavioural insights reveal anticipatory adaptations by pedestrians, such as slowing, stopping, or deviating, in response to perceived risk. The findings can inform more realistic simulation models, improve infrastructure design, and guide policy decisions aimed at enhancing safety at pedestrian-cyclist crossings. Future work should expand to cyclist behaviours and model calibration for improved predictive capacity.

This research aims to understand the influencing attributes of passengers and station layout elements on transfer walking times for metro transfers. Little research has been performed to include passenger-related attributes to model transfer walking time besides station layout elements. Through a literature review, the effect of gender, luggage size, group size and level of crowding are potential attributes. Furthermore, the vertical transport mode choice, including the lift, the waiting condition to board a vertical transport mode and the alighting location, are also part of the influential attributes besides the transfer length. The walking time and passenger characteristics have been collected through a covert observation. In the data analysis, the effect of group size, vertical transport mode choice, waiting condition to board and the alighting location significantly impact the transfer walking time and the walking time on a transfer segment level. These attributes have been captured in walking time and passing speed models for various transfer segment types. The walking time models can predict a lower, mean and upper bound of the walking time for each combination of attributes. The case study for the walking time collection was metro station Beurs, Rotterdam.

Predicting crowd movements in real-time during mass events has been shown to be a complex yet valuable task in order to reduce the risk of overcrowding. The aim of this research is to propose and validate a crowd movement forecasting method for which simulation is performed offline (i.e. prior to the event) but the forecast is done online, in real-time. A number of scenarios is formulated and simulated creating what is called a database of scenarios. In real-time, based on information from the event's crowd monitoring systems, a scenario from this database is then selected which corresponds to the prediction. The research is focused on addressing the concepts related to the two pillars of the method: the formulation of the scenarios to be included in the database, and the operationalization of the system to select a scenario in real-time.

The increasing passenger demand for rail service in the Netherlands, urges the Dutch infrastructure manager (IM) ProRail to increase network capacity. Instead of building new infrastructure, ProRail promotes information and communications technology solutions which aim for a more efficient utilization of the existing infrastructure. One of the ways to achieve improved capacity utilization is through decreasing the variability of train runs, i.e. by attaining more uniform train driving profiles. A driver advisory system (DAS) constitutes such a solution and it serves as a support for the driver to perform the train driving tasks. Additionally, the rollout of the European Train Control System Level 2 (ETCS L2) on the first part of the mainline Dutch network is planned for 2030. Developing a DAS compatible with ETCS L2 operation would yield high quality advice. Still, until the complete roll out of ETCS L2, ProRail aims to improve capacity on given bottlenecks using existing systems. A DAS constitutes a nonsafety critical, Grade of Automation 1 system. Under DAS operation, the driver adjusts the train controls and he/she is responsible for the safety of operations. A frequently arising problem when using a DAS, is that its advice usually leads to conflicts due to poor or no consideration of the actual traffic. An approach to handle this, is to provide the speed profile calculation module of a DAS with a dynamic speed profile that considers static and temporary speed restrictions as well as speed restrictions originating from the signalling system. The latter approach increases a DAS’s awareness regarding the actual signal state. This study aims to tackle conflicts with the latter approach. This approach for coping with conflicts when using a DAS is mentioned by several publications or commercial DASs but none of them explicitly defines how this is achieved. The main aim of this study is to increase the awareness of a conceptual CDASOn board regarding the actual signal state in order to minimise conflicts in disturbed operations. The proposed framework addresses the a CDASOn board operating on top of the Dutch signalling system NS’54 and the ClassB automatic train protection system ATBEG. In order to increase the proposed model’s effectiveness, this information must be provided in real time. Realtime signalling information is delivered on board by the maximum allowed speed data stream (Figure 1). It is proven that the only missing function from existing ProRail systems to provide this realtime information flow, is to determine the red signal in real time. Thus, the initial objective of this study can be scoped down to determining the red signal. It is also proven that the goal of determining the red signal is equivalent to the goal of locating the predecessor train. This information is planned to be fed to the on board equipment of a DAS via a novel data stream. Additionally, this section explained how this data stream fits to the train control architecture using a CDASOn board.

With the increasing popularity of active modes (mainly pedestrians and bicycles), the spaces shared amongst these modes are also rising. Such spaces are often referred to as shared spaces and can often be seen in public areas near train stations, shopping streets and educational institutions. Such spaces are said to enhance safety and resolve spatial limitations of the space while the users are seen to exhibit interesting and complex behaviours as they freely interact with different modes approaching from various directions. This creates a need to collect data and understand the behaviour of people within such spaces.
Many studies focusing on shared space interactions are limited by the current methods of data collection and data extraction. Other studies on people movement also face similar issues. Current data extraction approaches using cameras include repetitive and labour-intensive tasks which makes them costly and inefficient. Thus, this study focuses on automating the data extraction approach to obtain the ground-plane trajectories of people using the data recorded from a 3D-stereo vision camera. The data processing framework was divided into three main stages, (i) agent detection, (ii) ground-plane representation and (iii) agent tracking. The agent detection uses a neural-network-based detection model to detect people on the visual video which is then paired with the depth data to represent people on the ground plane. Each person is represented as a single point on the ground plane for every frame of the video. These points were then inputted into a tracking model to provides trajectories of people on the ground plane. Overall, this research integrates the recent state-of-the-art technologies into a single framework to automate the data extraction process and tests this framework in a real-world setting (behind the Amsterdam central station’s shared space area). For each stage, the advantages, challenge and recommendation for further improvements have been made. By automating the data processing of real-world datasets (including pedestrians and cyclists), This study reduces the cost and provides an easy way to collect and process data on real-world movements of pedestrians and cyclists. This will encourage urban planners to better understand people’s movement within their spaces and facilitate data-driven design approaches in the future.

At mass events, pedestrians can experience the level of crowdedness as unsafe, unpleasant and stressful. To gain a better understanding of perceived crowdedness, the effects of personal, trip and event characteristics at an event are researched. Data collection was performed by a simultaneous survey and monitoring study at the TT Festival in Assen and at the Red light district in Amsterdam. A SEM model shows that perceived crowdedness is influenced by mainly by the density, quantified using Wi-Fi sensor data. Besides that, trip purpose and familiarity with the event influence perceived crowdedness as well. Furthermore, this research shows that perception of safety, comfort, atmosphere and attractiveness of the environment are also related to the perception of crowdedness.