Annually, thousands of lives are lost to traffic accidents. To improve the safety of all traffic participants, the understanding and modelling of the limitations of human behaviour in traffic have continuously been researched. Currently, there is a lack of existing research on human predictions of other vehicles in traffic beyond binary decisions, such as whether the pedestrian will cross or whether another vehicle will accept the gap. This study conducted a human factors experiment with a novel response method where 30 participants viewed 168 unique scenarios for 5 seconds and then had to predict the intended direction the other vehicle would continue at the intersection. The direction predictions are a measure of how likely humans think the observed vehicle will go forward, left or right at the intersection. Analysis of the results showed that the heading angle and the relative position of the other vehicle had the greatest influence on the predicted direction and confidence of the response. Blinker use and deceleration had a lesser impact on prediction direction but significantly affected confidence. The lateral offset showed no statistical significance on the responses. The results highlight the limitations and inconsistencies in human predictions for other vehicles, particularly when the observed vehicle was positioned on the left or right side of an intersection, even when participants could focus solely on the other vehicle and no other distractions were present. Accounting for these inconsistencies when developing driving systems or testing autonomous vehicles can significantly enhance the safety and awareness of all traffic participants involved in an intersection.

With the advancements in automated driving, there is an increasing focus on passenger comfort and safety, fueled by the desire to establish a unique user experience identity among automotive companies. This study investigates the potential of the Galvanic Skin Response (GSR) as a physiological marker for assessing user experience. For this purpose, a test study of 32 participants was performed by collecting GSR measurements and self-reported comfort and safety scores, using a Wizard-ofOz setup on a closed test-track over repeated laps, alternating between distinct driving styles. Statistical analysis revealed the phasic maximum amplitude and peak count as the features most strongly correlating with both objective driving style and perceived comfort and safety ratings. The GSR measurements were also given as input for a predictive model for classifying the driving style, yielding an accuracy of 88.61%. General performance of the same model for perceived comfort and safety prediction on a five-point Likert scale was, however, notably lower, whereas participant-specific model calibration yielded substantially higher performance. This indicates that while the GSR consistently reflects physiological responses linked to perceived comfort and safety, the signal exhibits a strong inter-subject variability, highlighting the necessity of personalized calibration for accurate passenger-experience assessment.

This project aims to provide insights and a better understanding of power losses in cycling, with a particular focus on professional road cycling and the vibration losses experienced during rough road stages such as Paris–Roubaix. While aerodynamic drag and pure rolling resistance are well-documented, vibrational losses, caused by the excitation of the bicycle-rider system over uneven roads, are often overlooked or inadequately modeled. This thesis investigates vibration-induced losses up to speeds of 30 km/h using a controlled coast-down experiment over a simulated cobblestone surface, isolating their contribution from aerodynamic and classical rolling resistance forces.
The results show that vibration losses can reach up to 350 W and account for as much as 60% of total power losses under certain conditions. A clear quadratic trend with respect to speed was identified, challenging the conventional assumption that such losses can be included in the speed-independent rolling resistance term. Posture and tire pressure were also found to significantly influence these losses: riding out of the saddle consistently reduced vibration losses, while higher tire pressures led to greater total power loss due to increased vibrational dissipation. In contrast, rider behaviour, defined by varying levels of muscle contraction, did not yield consistent trends under the current experimental results.
To describe and better understand these dynamics, a simple empirical model was proposed, incorporating a quadratic dependence on speed, a linear dependence on pressure and a posture-dependent scaling factor. While intentionally limited in complexity, the model fits the observed data well and captures key interactions between the main influencing variables. These findings support the need to model vibration losses as a separate, speed-dependent component of total resistance. They also offer a useful starting point for future research focused on refining resistance models and enhancing energy efficiency in cycling.

Ferromagnetic materials in the walls and ground of buildings cause perturbations in the earth’s magnetic field. These spatial variations in the ambient magnetic field observed in indoor environments are mostly time-invariant. The deviations of the magnetic field can be captured in a magnetic field map, providing valuable information for indoor localization and navigation. Gaussian processes are a useful tool to model the ambient magnetic field, where the characteristics of the magnetic field are specified by the hyperparameters of the Gaussian process. To avoid the computational difficulties associated with full Gaussian process regression, a reduced-rank approximation is implemented. The magnetic field is measured using a magnetometer. A motion capture suit contains multiple magnetometers, whose relative positions are accurately determined. In this thesis it is researched if the quality of the magnetic field map can be improved when the measurement come from the motion capture suit. This is a challenging task, since the magnetometers are placed at different altitudes and the characteristics of the indoor magnetic field also vary with altitude. Consequently, the hyperparameters that best fit the measurement data are different per magnetometer. The negative effects of this observation are reduced by considering subsets of magnetometers operating at similar altitudes. However, an evident relation between the magnetometer’s altitude and its optimal hyperparameters has not been found. The hypothesis that complementing measurements from a single magnetometer with measurements of magnetometers operating at similar altitudes would improve the quality of the magnetic field is not supported based on the experiments. It is expected that either the magnetometer measurements lacked consistency, or that the assumption that the MOCAP suit returned the true magnetometer locations does not apply.

Riding a bicycle without hands, while being pleasant and sometimes challenging, exposes the self stabilizing properties of the bicycle. While human balancing techniques are widely understood, balancing on a bicycle is predominantly done by turning the handle bars and not many details about body motion control strategies are understood yet. Therefore, creating a rider model that controls the bicycle solely by lateral body motion enables these strategies to be amplified and better appreciated. Only controlling the bicycle by lateral body motion also gives more insight into the relation between roll and steering dynamics as the steering control is non-existent. Further inspiration for the model is obtained by analyzing motion capture data of a rider on a bicycle. This thesis describes the process of creating such a hands-free rider models and the development of simulation tasks to examine this control behavior. These tasks involve performing a double lane switch maneuver, a 90◦ turn and riding through perturbations. The model is created using the BRiM software package, and simulations are performed by optimal control program Opty, which allows for the symbolic equations of the Whipple bicycle model to be used for multi-body-dynamics calculations. After analyzing the control strategies employed by the rider, new insights are obtained on the lateral mechanics of bicycle control, paired with improved joint mechanics that have not been used before in bicycle-rider models. Such new joint mechanism modelling the seat connection between bicycle and rider, that couples lateral translations to body lean proves to be moreeffective than only rider body lean. Also, in terms of optimal control, using a multi-link pendulum model for the rider does not increase the control performance of the rider compared to a single pendulum rider lean model. Furthermore, novel evidence is presented that disproves the assumed need for counter-steering when attempting to make a turn. The rider can make a steady turning motion without having to initially steer in the opposite direction of the turn. Moreover, the effects of placing a spring between the front- and rear frame of the bicycle, both in stabilizing the bicycle as making the bicycle more responsive are investigated. From this it is found that adding a spring will increase the self-aligning properties of the front wheel but that it hampers maneuverability. Finally, the results of this thesis are put in perspective, and recommendations and improvements for future work are shared.

Automated driving is poised to transform the transportation landscape of the future, but several challenges remain before full automation is achieved. One of these challenges lies in managing perception uncertainties, such as those arising from radar and sensor measurements, while maintaining control in low tire-road friction conditions. These challenges often occur simultaneously in adverse weather conditions, but are typically researched separately. Their combined effect on safety and performance remains underexplored, even though addressing them together is critical for robust and reliable automated driving systems.

Lower tire-road friction limits the available tire force required for obstacle avoidance. Properly modeling these tire friction limits is particularly important in dynamic and uncertain environments to adequately account for the changing environment responsively. Moreover, addressing perception uncertainties and modeling low-friction conditions individually can significantly increase computational demands which poses challenges to achieve real-time performance required for real-world implementation. Therefore, this thesis jointly considers perception uncertainties and low tire-road friction conditions, accounting for their interacting effects. This is accomplished by addressing the following research question: "How can perception uncertainties be effectively integrated into motion-planning models for obstacle avoidance to improve performance and safety of automated vehicles in various road conditions?"

To address this question, a grid-based stochastic model predictive control framework is extended, implementing a non-linear bicycle model and a Fiala tire model (brush model) to consider realistic vehicle capabilities in low-friction conditions. Grid-based stochastic model predictive control reduces the uncertain obstacle environment into a set of linear constraints, utilizing an occupancy probability grid to effectively consider perception uncertainties. While the developed framework is presented as a proof-of-concept with a focus on safety and feasibility over real-time implementation, computational efficiency is not overlooked. By reformulating the constraints, the uncertainties are effectively accounted for while also reducing the optimization time by simplifying the probabilistic obstacle space to a deterministic convex region. If the nominal reformulation fails, a novel back-up method generates a conservative back-up set of constraints improving the safety and feasibility of the method. Two different back-up strategies are proposed, providing a trade-off between accuracy and computational effort.

To evaluate the contributions, three simulations were conducted. The first comparing the non-linear bicycle model and Fiala tire model to more simplistic models at various tire-road friction coefficients, highlighting the improvement in control of the proposed model in low-friction conditions. The second simulation evaluated the performance of the proposed back-up methods and environment representation by simulating tight scenarios that would fail using only the nominal approach. Feasibility rates increased compared to the baseline back-up method (43.8%) with feasibility rates of 62.5% for the precomputed back-up method and 75.0% for the current-state back-up method. These simulation results demonstrate that the capability of the proposed framework to compute feasible solutions and that the framework is able to compute valid hulls that are safe when the nominal approach fails to reformulate the constraints. The final simulation evaluated the performance of the complete proposed method by simulating both tight scenarios requiring a back-up method as well as various friction levels. These simulations demonstrated that the proposed framework is effective at lower friction levels, achieving high feasibility rates of 87.5% for the current-state back-up method, and 91.6% for the precomputed back-up method. This high accuracy is particularly promising for real-time applications using the precomputed back-up strategy, since this method leverages parallel computation of the back-up constraints.

This thesis demonstrated the effectiveness of the proposed extended grid-based stochastic model predictive control framework in considering perception uncertainties in low-friction road conditions. While presented as a proof-of-concept with an emphasis on feasibility and safety rather than real-time implementation, the proposed framework lays the groundwork for real-time applications, marking a step in solving all edge cases required to reach fully automated vehicles.

The advancements in automated driving hold the potential to improve our daily lives by reducing traffic accidents, minimizing congestion, and allowing more time to spend on other activities. However, realizing these benefits depends heavily on widespread public acceptance, which, according to current research, remains low. Trust and comfort have been indicated as key factors to increase acceptance, highlighting the importance of finding methods of measuring these concepts. This thesis contributes to the development of objective methods for evaluating the emotional state of drivers and passengers in automated vehicles.
In this thesis, an experiment was designed and conducted using a driving simulator to allow participants to experience a ride in an automated vehicle. The aim was to elicit varying levels of comfort by altering the driving style and introducing the presence of a pedestrian, while simultaneously collecting a comprehensive dataset to analyse these comfort levels. This dataset includes continuous subjective comfort ratings given by the participant, vehicle dynamics from the real-world drive on which the simulation was based, webcam footage monitoring the person’s facial expression, Galvanic Skin Response, heart rate (variability), and eye-tracking from 32 participants. Such comprehensive datasets are rare in the literature and provide valuable opportunities for future research to compare different signals and explore their interrelations. From the subjective comfort ratings, it was found that the driving style was a bigger factor than the presence of a pedestrian. Even though the pedestrian did cause a decrease in comfort, the difference between the two driving styles was found to be significantly bigger. For facial expression recognition, a state-of-the-art model was successfully implemented. With minimal lighting conditions, the face could always be detected, and expressions were successfully classified with corresponding emotion labels from the universal set of emotions. Out of the 32 participants, 24 were included in the analyses. Most of these participants (15/24) did not show any detectable reaction in their facial expressions to the critical event. Amongst the 9 participants who did, 8 of them showed a Happy expression, and only 4 showed a Surprise expression. Fear was never dominant. This result shows that, in the current experiment, Facial Expression Recognition is not a reliable method for discomfort detection in Automated Vehicles. Additionally, a neural network was implemented to predict a person’s subjective comfort based on vehicle dynamics and their Galvanic Skin Response (GSR). The model was validated using Leave-One-Out Cross-Validation (LOOCV), where one participant was excluded from the training set and their data was used for testing. The results were promising, as the self-reported comfort and the model-predicted comfort showed a positive correlation across all participants. These findings demonstrate the potential for objective comfort assessment in automated vehicles, reducing the biases inherent in subjective evaluations and paving the way for further research in this field.

People need to have a comfortable experience in vehicles nowadays. However, they are continuously exposed to vibrations from the vehicle. Madymo active human models (AHM) can be used for comfort analysis and to learn how vibrations influence the human body in several aspects. However, existing AHM are very time-consuming due to their complexity, and the correspondence with human comfort data is only reasonable. In order to more effectively analyze motion comfort, a computationally efficient simplified human model (SHM) is developed and validated. The human body model has 36 degrees of freedom (DoF) considering the following segments: pelvis, two thighs, two lower legs, two feet, lower torso, upper torso and head. The model is validated in fore-aft, lateral and vertical vibrations. The model's postural stabilization parameters are tuned manually, by gradient search and grid search in sequence. Manually tuning gives a group of initial values of parameters for further optimization. According to the results, failure of the gradient search illustrates that this optimization problem is non-smooth. At the same time, grid search gives a relatively better result but also shows that the current cost function does not perfectly represent a good fit and needs improvement. A comparison between AHM and the SHM shows that the SHM has a similar or even better fit for most signals while being a largest factor 116 faster. A comparison of parameters between SHM and multibody human models in the literature shows that the structure of models affects the values of the same parameters. A comparison of different time steps illustrates that shorter time step does not necessarily give higher accuracy for this SHM. Suggestions such as adding muscles and changing body shapes are given for further improvement.