Optimisation of public transport networks are crucial for a well-functioning city or a large urban agglomeration. Public transport is the most efficient way for large groups of people to travel in and to a city. In this report the optimisation is confined to the network optimisation of the bus. Based on the available budget choices have to be made for the network design to maximise the ridership.
The focus of this thesis is the stop and line spacing of the bus, the distance between sequential bus stop and parallel bus lines respectively. Changing the stop and line spacing have an effect on the running time of bus users, with a trade-off between the walking and in-vehicle time. A self-designed total travel time model optimises the weighted total travel time with weights on the walking time and frequency. The frequency here is based on a combination of stop and line spacing. The stop and line spacing affect the running time and thus how much a vehicle serves a route per hour. This in turn determines the frequency. The result of the total travel time model is an optimal stop spacing of around 540 meter and a line spacing of 700 meter. Analytical models by other researchers found an optimum stop spacing in between 600 and 650 meter and a line spacing of 750 meter. In practice guidelines are used where the stop spacing is around 400 meter and the line spacing is around 550 meter. The result of the self-designed model is different because the walking weight is exponential and the frequency weight is tied to the running time of the bus. The conclusion is that the optimal stop spacing is higher than is mostly applied in bus networks. A higher stop spacing means a higher average speed and a lower travel time in the bus. The downside of a higher stop spacing is that the walking distance increases which effects the ridership. This is the reason the self-designed total travel time model has a larger focus on the walking distance, and has resulted in a lower optimal stop and line spacing than in analytical models of other researchers.
The effect the walking distance has on the use of the bus is different in different area types in an urban agglomeration. A regression analysis on the relation between the stop spacing and sociodemographic characteristics has been performed for the analysis of this effect. The data used for the regression analysis is gathered for the city of Rotterdam and the surrounding towns. For an area with a high population density, income, and car ownership in combination with a large distance, around 10 km, from the city center a stop spacing of 600 to 700 meter is recommended. The lower the distance to the city center, the lower the stop spacing, and thus for a similar area type around 5 km from the city center a stop spacing of 500 to 550 meter is recommended. For an area with an average population density, income and car ownership in combination with a high distance (10 km) to the city center a stop spacing of 475 to 525 meter and in combination with a lower distance (5 km) a stop spacing of 450 to 475 meter is recommended. The reason why these values are lower than for the first area type is because there are more activity facilities. This means that there are more potential destinations in this area type and for users a stop close to a destination is important for the choice to use public transport. The larger the area of these facilities, and the higher the number of facilities, the lower the stop spacing. In the city center the recommended stop spacing is therefore 425 to 450 meter. The exception to this is the area close to the central station of Rotterdam, here the stop spacing is higher (550 to 600 meter) because close to a station people are not going to use the bus, but the train which has a higher operation speed and thus is a higher quality mode. For an area with a low population density, income and car ownership a stop spacing of 550 to 650 meter is recommended. In this area type there are a lot of captive riders, who are dependent on public transport and are willing to walk further than other types of public transport users. For this group it is important that bus stops are close to activity centers. If the stop spacing in this area becomes too high the number of trips made by captives decreases, even if the number of users stays the same. Having bus stops close to destinations compensate for the higher stop spacing. These destinations could also be a train, metro and/or tram stop. This complies with the higher willingness to walk, and makes the bus network more efficient.
For the line spacing it is more complicated to recommend values for certain area types. However recommendations are given to the network type and design, which is closely related to the line spacing. Because different area types have different characteristics a hybrid network is the most effective solution. In the city center a grid network is used to distribute users equally. Radial lines are used to connect areas outside the city center with the city center and ring lines are used to create connections between these areas if the demand for this is there. The further away from the city center the lower the bus stop density.

Background
The steady development of automated vehicles (AVs) promises significant benefits in terms of traffic safety and efficiency. However, the transition to fully AVs and their deployment on the road will be gradual, leading to a phase of mixed-traffic conditions where AVs at various levels coexist with human-driven vehicles (HDVs). This transition poses unprecedented hurdles, requiring a deeper understanding of the emerging challenges for AVs in sensing and perceiving road environments, as well as in the novel interactions between AVs and HDVs. Furthermore, the social compliance of AVs and the optimization of their deployment strategies need to be considered as well.

Contents of this Thesis
This thesis addresses the multifaceted challenges associated with AVs’ development and deployment in mixed-traffic environments. The main objective of this thesis is to enhance the capabilities of AVs enabling them with a wider Operational Design Domain (ODD) and thus facilitate the implementation of safe, efficient, and socially compliant automated driving in mixed traffic. Referring to the modular design of AV systems, three key perspectives, i.e., sensing and perception, anomaly detection, as well as planning and control, are tackled in this thesis. To be specific:
Chapters 2-4 focus on enhancing sensing and perception capabilities through the development of hybrid spatial-temporal deep learning models and self-supervised pretraining methods. Lane detection is chosen as the focus of these chapters since it is vital for current vehicle localization and positioning, and it is also the foundation of various automated driving features. The main findings of these chapters are summarized as follows.
Chapter 2 presents a pioneering hybrid spatial-temporal sequence-to-one deep learning architecture tailored for vision-based lane detection tasks. By integrating the spatial convolutional neural network (SCNN) with spatial-temporal Recurrent Neural Network (RNN) modules, this architecture effectively captures correlations and dependencies among continuous image frames. Through extensive experimentation on various driving scenes, including challenging scenarios, the proposed model variants exhibit superior performance over existing state-of-the-art models. Notably, even the lighter model variants demonstrate remarkable accuracy, outperforming their counterparts while maintaining lower computational complexity.
Building upon the foundation laid in Chapter 2, Chapter 3 focuses on refining vision-based sensing and perception through the development of customized spatial-temporal attention mechanisms. These mechanisms, including temporal attention, spatial-temporal attention, and spatial-temporal attention with fully connected layers, are meticulously designed to optimize the utilization of spatial-temporal correlations across different regions of interest within the consecutive image frames. Leveraging linear Long Short Term Memory (LSTM) neural networks in conjunction with the proposed attention blocks, this chapter demonstrates the feasibility of lightweight and computationally efficient solutions for sequential deep neural networks (DNNs). Through rigorous experimentation, ablation studies, and comparative analysis across diverse datasets, the effectiveness of the proposed attention mechanisms in enhancing lane detection performance is convincingly established.
In Chapter 4, the exploration of enhancing vision-based sensing and perception capabilities continues with the introduction of a self-supervised pretraining method employing masked sequential autoencoders (MSAE). This innovative approach leverages both labelled and unlabelled data to improve detection accuracy and expedite the training process of DNN models dedicated to lane detection tasks. Additionally, a customized Focal Loss based PolyLoss is introduced to further enhance the detection accuracy. Through comprehensive experimentation and comparative analysis, the efficacy of the proposed pretraining method and loss function is demonstrated, showcasing substantial improvements in lane detection performance across diverse driving scenarios. Specifically, the utilization of MSAE-based pretraining and the adoption of the customized PolyLoss result in superior performance metrics, underscoring the pivotal role of self-supervised learning techniques and tailored loss functions in fortifying the robustness and efficiency of vision-based sensing and perception systems in AVs.
These chapters address the challenges of vision-based lane detection, crucial for AV navigation and safety.

Chapters 5-6 delve into anomaly detection, investigating techniques for identifying abnormal lane rendering in digital map applications and detecting anomalies in driving behaviour.
Chapter 5 introduces an innovative approach to anomaly detection in lane rendering images of digital map applications, utilizing Transformer-based models with self-supervised pretraining and customized fine-tuning. By transforming anomaly detection into a classification problem, the chapter proposes a four-phase pipeline that includes data pre-processing, self-supervised pre-training with masked image modelling (MiM), customized fine-tuning using cross-entropy-based loss, and post-processing. Experimental results demonstrate the pipeline’s effectiveness, with significant improvements in detection accuracy and reduced training time achieved through self-supervised pre-training. Ablation studies regarding tackling the problem with different numbers of classes further validate the pipeline’s performance enhancements, particularly in addressing data imbalance. This approach not only enhances anomaly detection accuracy but also contributes to reducing labour costs associated with manual labelling and anomaly detection efforts, offering significant societal benefits.
Additionally, Chapter 6 explores the critical task of detecting abnormal driving behaviour, addressing the need for more feasible and efficient approaches by leveraging semi-supervised ML methods. Utilizing large-scale real-world driving data, the study develops a semi-supervised ML model based on Hierarchical Extreme Learning Machines (HELM). This approach utilizes partly labelled data and introduces Surrogate Safety Measures (SSMs) (specifically the event-baed safety indicators of Two-Dimensional Time-To-Collision (2D-TTC)) as the pivotal input features to enhance performance. Results demonstrate the effectiveness of the proposed semi-supervised ML model, showcasing superior performance compared to baseline methods. The integration of SSMs significantly improves detection accuracy, highlighting their significant role in enhancing model performance. By leveraging unlabelled data for training and only a small sample of labelled data for fine-tuning, the proposed semi-supervised approach achieves competitive performance while reducing dependency on fully labelled datasets, making it suitable for real-world applications.
To sum up, the exploration of semi-supervised and self-supervised ML methods presents promising avenues in anomaly detection. The pioneering research presented in this thesis represents a significant stride towards leveraging data-driven ML-based anomaly detection methodologies to enhance the safety of driving.

Chapters 7-9 shift the focus to planning and control strategies for AVs, presenting a comprehensive examination of decision-making frameworks and control algorithms. These chapters introduce a conceptual framework aimed at fostering socially compliant driving behaviour and propose a range of model-based and learning-based approaches.
Chapter 7 lays the groundwork by introducing a conceptual framework that emphasizes socially compliant automated driving. This framework encompasses various social components such as cultural nuances, norms, and driving styles. A key innovation is the introduction of bidirectional behavioural adaptation, highlighting the dynamic interactions between AVs and human drivers. Furthermore, the framework advocates for the incorporation of a spatial-temporal memory module to enable continuous refinement of driving strategies, thereby promoting adaptability and safety in diverse traffic scenarios. Validation through an online expert survey lends credence to the framework’s efficacy. This conceptual framework lays a solid foundation for learning-based and model-based approaches for implementing planning and control algorithms for automated driving.
In the learning-based approach explored in Chapter 8, Deep Reinforcement Learning (DRL) takes centre stage, with a focus on integrating safety, efficiency, comfort level, and energy consumption considerations into the learning framework. Multiple DRL algorithms are evaluated across diverse driving manoeuvres, particularly roundabout driving, highlighting the importance of real-world requirements in reward function design and simulation-based training. Among the compared DRL algorithms, Trust Region Policy Optimization (TRPO) emerges as leading in safety and efficiency, while Proximal Policy Optimization (PPO) excels in comfort during roundabout driving. Moreover, the extension of the training environment to encompass various driving scenarios showcases the adaptability of DRL models to train a uniform driving model for real traffic environments, signalling promising avenues for future research.
Regarding the model-based approach, Chapter 9 introduces the DRF-SVO-MPCC algorithm, aimed at enhancing AVs’ understandability and predictability to human drivers, particularly during interactions with HDVs when driving through the roundabouts, as this challenging manoeuvre involves large curvature and tackles both longitudinal and lateral control. This algorithm integrates the perceived Driving Risk Field (DRF), Social Value Orientation (SVO), and Model Predictive Contouring Control (MPCC), enabling AVs to navigate social scenarios with sensitivity to the welfare of surrounding HDVs. Simulation experiments, conducted on various roundabout scenarios, underscore the algorithm’s superiority in trajectory tracking and adaptability to different driving styles, ensuring safety and social compliance. The findings illuminate the potential of the DRF-SVO-MPCC algorithm in fostering harmonious interactions between AVs and HDVs, setting a precedent for socially aware automated driving systems.

Overall, this thesis represents a solid endeavour to advance the planning and control capabilities of AVs in mixed-traffic environments. Through the development of novel conceptual frameworks and innovative model-based and learning-based algorithmic solutions, it lays the groundwork for the realization of safe, efficient, socially compliant, and adaptable automated driving, contributing to safer and more harmonious transportation systems.

Conclusion and perspectives
In summary, this thesis contributes to advancing the knowledge of how to improve automated driving systems in the realms of sensing and perception, anomaly detection, as well as planning and control. By integrating theoretical frameworks, methodological innovations, and data-driven empirical evaluations, notable progress has been achieved in fostering the development of safe, efficient, and socially compliant automated driving within mixed-traffic environments.
Despite the considerable progress made, several directions for future research have been identified. These include the imperative for more expansive high-quality datasets, exploration of domain adaptation techniques for both sensing and anomaly detection tasks, as well as the seamless integration of model-based and learning-based methodologies for planning and control. Additionally, transitioning towards a unified driving model and effectively addressing the complexities of multi-agent interactions in intricate urban settings remain pivotal areas for further exploration. Furthermore, interdisciplinary collaboration will be instrumental in harnessing the full potential of automated vehicles to revolutionize transportation systems.

The deployment of moving sensor platforms (e.g., self-driving cars, drones, and other instances) with advanced sensing is rapidly increasing the capture of human features at unprecedented temporal and spatial scales, especially in cities. This thesis advances knowledge on extracting information from this novel data source for traffic research and practice, while highlighting implications for privacy and beyond. Findings provide insights for traffic control and management, policy development, and anyone involved in responsible urban innovation.

Private cars are inefficient in terms of land allocation. Car-free neighbourhoods offer a solution to this inefficiency by enforcing less car-centric urban design, expensive parking fees, and prioritising more sustainable and space-efficient modes of transportation. Carsharing services emerge as an alternative to private car ownership by providing vehicles available for short-term use. Notably, not much is known regarding carsharing services as an alternative mobility solution in car-free neighbourhoods. The present thesis investigates pricing strategies that could be implemented for business-to-consumer carsharing services in car-free residential neighbourhoods to align with the diverse needs and preferences of potential users. To this end, this thesis employs a case study of the car-free neighbourhood Merwede. Specifically, a cost-benefit analysis is conducted to estimate the potential value of carsharing in a car-free context. Emphasis is put on policies that may stimulate carsharing adoption and on determining equitable service pricing strategies.  To predict the modal split of future Merwede residents, datasets – containing mobility information for Utrecht residents clustered by income group – are used. To determine the influence of a service price change on carsharing demand, per-income-group price elasticities are estimated. This resulted in the following values for low-, middle-low-, middle-high-, and high-income groups, respectively: -0.8, -0.5, 0.4, and -0.1. Three stakeholders are considered in the analysis: carsharing service operators, users, and the Municipality of Utrecht. Results indicate that subsidies for lower-income residents in the form of a trip credit incentive (of 50 or 100 euros per month per user) yield positive total net present values and higher carsharing adoption rates. This thesis concludes by stating that while the generalisability of these findings may be limited, the CBA model synthesised for Merwede provides a case study from which future car-free neighbourhoods – aiming to implement carsharing – may learn.

The transport sector needs to change to meet some of the global challenges we face: climate change, population growth and urbanisation. It needs to become more sustainable, which is why the Netherlands wants to introduce more high-quality bus systems. However, there are some barriers to implementing these systems.

In this thesis, through a literature review and interviews with Dutch stakeholders, a definition of the high-quality bus system is found, including its main characteristics. The most important part of this thesis is to map the existing problems. In order to solve these problems, a roadmap is designed to help with the implementation by breaking the process down into smaller, more manageable steps.

A high-quality bus system should above all be fast, frequent, and reliable, but comfort is also important for such a system. The most commonly used terms are Bus Rapid Transit (BRT) and Bus with a High Level of Service (BHLS). The literature and the interviews confirm that there are still many problems that should be addressed in the roadmap, from cooperation problems to short-term thinking by politicians, from blind commitment to rail to underestimating the social value of public transport.

The proposed roadmap consists of eight blocks. It begins with the creation of a city-wide vision for mobility and the formulation of a programme of requirements for the entire public transport system (i). These are translated into objectives and a programme of requirements for the line to be designed (ii). In addition, cooperation with all parties should be well regulated (iii). The next important step is to analyse the target group of (potential) passengers, the stops and the bottlenecks and intersections (iv). Only after that is the system choice made (v). If a high-quality bus system is chosen, it needs to be designed (vi). Among other things, the route, the operation, the first/last mile transport and the branding have to be designed. When the design is ready, the line can be built, and operation can begin (vii). This can be done in stages, addressing the main bottlenecks first and the rest later. This is followed by the often overlooked evaluation, which is important for learning lessons for future projects.

Stakeholder feedback and a case study show the value of the roadmap for implementation in the Netherlands. By choosing the modality later in the process, compromises in quality are avoided. In addition, the in-depth analyses improve the quality of the line and enable faster implementation by identifying the main bottlenecks. The main value of the roadmap lies in its integral approach, rather than an approach from one perspective. This means that the problems of all stakeholders can be addressed to design a high-quality public transport system that works for all.

This research provides insight into ways to increase accessibility for Planned Special Events (PSEs) via discrete choice modelling. A special focus within this study is put into the AFAS AZ Stadium in Alkmaar. During the events held at the AFAS AZ Stadium disturbances are experienced by its visitors with regard to crowding levels and access/egress times. Quantifying the traffic volume on the infrastructure around the venue is done via revealed preference methods. Cyclists are counted via pneumatic tubes. Whereas pedestrians are measured via radio-wave sensors. To find ways to mitigate these problems, interviews are held with organisers of similar PSEs within The Netherlands.

To find the effectiveness of the obtained accessibility measures a stated preference study is held. From this study it is obtained that the transportation mode habit that one has is a key factor within the mode choice. Besides the effect on accessibility additional decision factors are discussed to find the best measure which can be applied best at the AFAS AZ Stadium. Based on the decision factors and the performance of the accessibility measures a total of six effective measures are determined.

This dissertation discusses how mode and route choice behavior might change with the introduction of future transportation modes when potential users are unfamiliar with such systems. It uses discrete choice models and supernetwork models without mode-specific constants and parameters with revealed preference data of current modes to understand how future modes could impact mobility effects such as mode choice, travel times, and resistance.

Road traffic safety is a pressing global concern, with millions of yearly fatalities and injuries. This study aims to address the detection of abnormal driving behaviour. Traditional supervised approaches face limitations due to the need for labelled abnormal driving data. To overcome this challenge, semi-supervised machine learning models are explored and developed in this research.
Machine learning is utilized for abnormal driving behaviour detection because it offers a data-driven approach that adapts to different scenarios and captures subtle patterns. Furthermore, its scalability allows for efficient analysis of large datasets, leading to accurate identification of abnormal driving behaviour and valuable insights for enhancing road safety measures. Most existing machine learning (ML) based abnormal driving detectors rely on (fully) supervised ML methods, which require substantial labelled data. However, in the real world, labels are only sometimes available, and labelling large amounts of data is tedious. Thus, there is a need to employ unsupervised or semi-supervised methods to make the detection process more feasible and efficient. Luckily, it is possible with the advent of deep neural networks, especially autoencoder-based ones. This thesis develops and compares three ML methods: supervised (e.g. XGBoost and Random Forest), unsupervised ML (e.g. Isolation Forest and Robust Covariance), and semi-supervised ML (Hierarchical Extreme Learning Machines). Comparison results show that the semi-supervised deep learning model outperforms unsupervised methods exhibiting higher prediction accuracy and delivering acceptable results compared to the fully supervised models.
Moreover, previous ML-based approaches predominantly utilize basic car motion features (such as velocity and acceleration) to label and predict abnormal driving behaviours. In contrast, this thesis introduces Surrogate Measures of Safety (SMOS) as features for ML models to identify abnormal driving behaviour.
The results indicate that the supervised model performs best under the same conditions. However, relying on a large amount of labelled data in supervised models can pose challenges in real-life scenarios or when dealing with massive datasets. The study highlights the significance of Surrogate Measures of Safety (SMOS) and demonstrates the potential of HELM in effectively identifying abnormal driving behaviour. The introduction of SMOS significantly improves the performance of both unsupervised and semi-supervised models. The unsupervised model shows the most substantial improvement, increasing accuracy from less than 50% to over 90%.
While the Isolation Forest and Robust Covariance models fail to detect abnormal driving behaviour without including SMOS, the semi-supervised HELM model exhibits promising results even without SMOS. However, further research is necessary to address limitations and enhance the findings. While valuable, the current dataset used in this study may only encompass some types of abnormal driving behaviour. Future research should incorporate a more diverse dataset that covers a broader range of abnormal driving behaviours. The analysis should include multiple SMOS features, such as Post Encroachment Time (PET), to comprehensively understand abnormal driving behaviour and improve safety measures.

- There is increasing attention in the Netherlands in the topic of transport
poverty and accessibility, with several publications discussing the need for accessibility standards to indicate injustice in the transportation system. Numerous case studies can be found where accessibility has been measured and assessed for fairness in the transport system, assuming that low-income households rely on public transport. This research reveals that up to now, the accessibility by public transport for low-income households and the unemployed is overestimated. Transport costs do not only have a diminishing effect on the accessibility by car, but also limits the accessibility by public transport. By means of the methodology ’Designing fair transportation systems’ it was possible to evaluate the job accessibility in the Parkstad region, a region where income on average is the lowest in the Netherlands and the unemployment rates the highest. The limited job accessibility by both car and public transport raises the question to what extent transport poverty contributes to the high unemployment rates in this region. Municipalities are recommended to use these results to further explore what the population groups suffering from transport poverty need and propose interventions to improve job accessibility for those who need this the most.

The implementation of a bus system is an attractive alternative for decision makers since it has a short lead time and the service can be easily adapted depending on the demand. Due to the flexible nature it allows for the use of fulfilling different magnitudes of demand. Usually bus service provide a lower level quality service, however busses can be configured in such a way that they also provide higher level of service up until a point where they compete with rail systems. In the Netherlands referred to as HOV (“Hoogwaardig Openbaar Vervoer”, high-level public transport). This is a large spectrum with many different configurations. One should still consider the need for a HOV bus service as it comes with extra investment costs compared with the regular bus. Therefore, this research was looking into where there is no difference in valuation between a regular bus line and a higher level of service bus line. Through interviews, analysis of current lines and looking into what decision makers promise as a higher level of service attributes could be defined which could be seen as characteristic for a higher level service bus service. Additionally, the commuters perspective played a important role as most literature mostly considered only the decision makers perspective. A method has been demonstrated which shows how commuter needs can be used as and translated to operational characteristics with which decision makers specify, plan and design current and future higher level of service bus lines. The method includes a MNL model which expresses the valuation of the characteristics attributes of high level bus services which allows for the prediction of the modal split between a bus service and the current commute. This gave insight on which attributes a differently valued between a regular bus and a HOV bus and when a HOV bus service is equally valued as a regular bus service, hence not justifying investments in a higher level service. It was found that in most cases the HOV bus service results in a higher modal split despite of having a lower valuation for most of the characteristic attributes. The commuter seems to have more trust in the reliability promoted by higher level bus services.

This PhD thesis delves into the intricate realm of supply-side behavioural dynamics and operations within the ride-sourcing system. It investigates drivers' pivotal decisions, focusing on ride acceptance and relocation choices at the operational level. By leveraging a unique dataset, novel theories on ride-sourcing drivers’ behaviour are developed and thoughtfully integrated into a system operation model, shedding light on their far-reaching impacts on the overall ride-sourcing system performance.

Dedicated lanes have been proposed as a potential scenario for the deployment of connected and automated vehicles (CAVs) on the road network. However, knowledge on the design and operation of DLs and their impacts on the behaviour of drivers of CAVs and manual vehicles is lacking in the literature. This dissertation provides a research agenda on design and operation of dedicated lanes and investigates the impacts of such lanes on the behaviour of human drivers.

Travelers do not like transfers in their Public Transport journey; it gives a disutility. Minimizing this disutility is valuable to increase travelers' satisfaction. To be able to reduce this disutility, transfers have to be identified first. Multi-operator smart card datasets allow for the identification of transfers between different operators, as well as between the same operators. However, it is unknown in the literature how such a multi-operator smart card dataset can contribute to the minimization of transfer disutility in Public Transport. An answer is given by analyzing a multi-operator smart card dataset for the Haaglanden area in the Netherlands that uses a 35-minute time interval to identify transfers. Also, a measure has been implemented for one of the transfers with the highest potential for transfer time loss minimization and the effects on the network are examined in a transport model. It is found that a multi-operator smart card dataset can identify important transfer stations and individual transfers, for which the associated disutility factors can then be identified manually. Then, measures can be implemented to reduce the disutility of a transfer. The measure, a reduction of the waiting time, implemented on a transfer in this study resulted in changing traveler's route- and mode choices. For further research, it is recommended to explore the possible effects of implementing the measure as it can lead to crowding or other inefficient transfers, which could increase the disutility. Furthermore, further research should look into the disutility values of the case study to draw better conclusions whether, and to what degree, (dis)utility factors explain transfer flow sizes for this case specifically.

One of the current challenges withholding personalised lane-level driving advice is the inaccuracy and error of GPS signal from commonly used navigation devices and mobile phones. These GPS signals have an uncertainty margin up to several meters, therefore potentially indicating the vehicle location on a different lane than the actual lane it would be in. This unreliability therefore currently makes it impossible to accurately recognise lane changes from solely this data. This study looks into the recognition of lane changes from only Floating Car Data by the use of a Random Forest algorithm. In order to find the ground truth, a trajectory reconstruction algorithm is implemented, which uses the matching of trajectories with loop detector passages in order to find the lane a vehicle is in at each loop detector location. This information is then used to know whether, for each vehicle, a lane change is made on the road section in-between two consecutive loop detector locations. By training the model on this data, it was found that when using solely Floating Car Data, lane changes can be recognised with an accuracy of up to 64%. Indicators for lane change were found to be the lateral distance of a vehicle to the middle of the road, as well as the heading of the vehicle. The study additionally looks into a rule based method of lane change recognition, which is compared with the Random Forest model.

The research of connected automated vehicles (CAVs) is an emerging topic within the field of transport & planning. It is not a question of whether the vehicles will be available for commercial use, but rather a question of when they will arrive. The safety of these vehicles is a necessary and ongoing discussion. In current research, a consensus is reached that crashes occurmainly due to human error. This human error is either due to negligence of the driving activity, like drunk driving or texting while driving, or due to incorrect decision making at critical moments. This study focuses on the topic of traffic safety with the principle of herd immunity in mind. It is theorised that crash risk can behave similarly to how a virus behaves (where crash risk is the chance that a crash occurs at a certain point in time). It spreads and infects vulnerable members of the population. The aim of this study is to determine whether the principle of herd immunity can be applied to car traffic, and if so, to what extent they can be compared through an impact assessment. The research question attached to this is: "How is traffic safety influenced by connected (automated) vehicles considering the concept of herd immunity?"

Advanced Driver Assistance Systems (ADAS) are becoming more available and will become mandatory for all new vehicle models from 2022 onward. In order to achieve the highest safety benefits, it is important that these systems are available. Lane Keep Assist (LKA) is part of ADAS and assists the driver in the lateral control of the vehicle. Lane markings are used by both human drivers and machine vision to stay on the road, but factors contributing to lane marking detection in different driving conditions are mostly unknown. A field test was conducted on Dutch provincial roads to evaluate lane marking visibility properties in relation to the LKA detection performance of different sensor types. The LKA detection performance of the mono camera was found to be higher in most weather and illumination conditions than the detection performance of the mono camera with infrared. The mono camera with infrared had a higher detection performance during rain in nighttime conditions than during dry daytime conditions. The highest detection performance for the mono camera and the mono camera with infrared were 97% in dry nighttime conditions and 91,4% in sunset conditions, respectively. Binary logistic regression was used to determine the effect of lane marking properties on the lane detection performance. A profiled lane marking type was found to increase the detection likelihood by 6-8 times as opposed to a smooth lane marking type. Other visibility properties, such as retroreflectivity and contrast with the road surface, were not found to be a significant contributor to the detection performance.

The goals of bus services can be split into patronage (regarding the number of people that actually use the service) and coverage (regarding the number of people that are able to use the service) goals. Current bus networks are designed primarily for the coverage goals, as a consequence the patronage is lacking resulting in only few bus services to be cost-efficient. Literature provides several practices for increasing the cost-efficiency, but the effects on the coverage function are often neglected. This report follows a design process for a Dutch case in order to find how the coverage can be enhanced by including several new practices from the beginning of a network design. New design tools were
developed in order to compare four different network designs relying on various coverage practices. The results show that a two layered system consisting of a high quality service combined with a supporting service in the form of a regular and/or demand responsive service is the best approach. Within the existing cost constraints, it is possible to create a large high quality network that relies on bicycles as access mode. By including shared bicycles, the coverage function is ensured while increasing the cost-efficiency of the network by allowing for faster routing. As a result, 10 percent more trips were made per timetable hour and 3 percent more passenger kilometres were covered. Replacing regular fixed line services by demand responsive services resulting in only a marginal increase that was highly dependent on the costs not to turn out higher. In combination with bicycle sharing, the results turned out to be much more positive with an additional 11 percent increase. By using a two layered system a high quality service can be provided that also serves a large coverage function. Of the supporting services, the use of shared bicycles allow for opportunities to improve even further, especially when combined with demand responsive transport.

Driving involves communicative interactions where human drivers use communication signals to negotiate their right-of-way for road safety. The introduction of automated vehicles (AVs) in mixed-traffic environment, where human drivers will interact with AVs, will affect the nature of these communicative interactions. AVs and human-driven vehicles (HDVs) use different communication forms (e.g., vehicle-to-vehicle communication between AVs vs eye-contact between humans). This raises road safety concerns. AVs might need a communication system that conveys intent to HDVs, clearly. This research investigates the potential of external human machine interfaces (eHMIs) in AV-HDV interactions at T-intersections to improve communicative interactions. Traditional traffic signals were used as inspiration to develop two eHMIs concepts, one placed on vehicle while the other on infrastructure. eHMI on vehicle might require less visual scanning from the interacting human to know AV intent and speed, but on the other hand, the AV might not be in driver field-of-vision compared to an eHMI on the infrastructure. The effects of eHMIs were investigated using a driving simulator with forty-six participants. The results show that both eHMIs had a significant and positive effect on driver trust, acceptance, and emotions. Drivers were calmer with eHMI placed on the infrastructure than on the vehicle. Both eHMIs reduced the time to maximum braking of human drivers and increased their compliance with AVs. The eHMI on vehicle reduced critical interactions, measured by the Post Encroachment Time, between AVs and HDVs. It is concluded that eHMIs can improve AV-HDV communicative interactions at T-intersections. No significant differences were observed between the eHMI conditions in participants’ preference and efficiency of the AV-HDV interactions, measured by human driver compliance. Hence, this research recommends further investigation of eHMIs in different on-road interactions.

This dissertation aims to understand the behavior of pedestrians and cyclists when interacting with automated vehicles (AVs). The role of AVs’ characteristics such as their physical appearance, whether a driver is present, the presence of external communication interfaces, and factors pertaining to the behavior of the vehicle were investigated using virtual reality road crossing experiments. In addition, psychological factors that could be affected by the presence of AVs were included.

Reducing congestion through persuasive in-car advice.