Balance control relies on the integration of visual, proprioceptive and vestibular information, with the relative contribution of these sensory signals depending on task demands. Although vestibular contributions to standing and walking have been extensively investigated, their role in cycling balance remains largely unknown. This study used electrical vestibular stimulation (EVS) to introduce a vestibular-driven error signal and investigate how vestibular perturbations influence balance-related kinematics and upper-body muscle activity during seated cycling and whether these responses are modulated by cycling speed. Sixteen healthy young adults performed non-pedalling treadmill cycling trials at 5, 10 and 18 km/h under both non-EVS and stochastic EVS conditions. Steer and roll kinematics and activity of six upper-body muscles involved in steering and postural stabilization were recorded. Time-domain measures quantified changes in response magnitude, while gain and coherence between EVS and steering rate, roll rate and muscle activity were calculated in the frequency domain to characterize vestibular coupling. EVS significantly increased root mean square (RMS) steering rate and RMS roll rate and increased mean activity in all investigated muscles except the erector spinae. RMS steering rate and EVS-steering gain decreased significantly with increasing cycling speed, whereas RMS roll rate and EVS-roll gain were not significantly affected by speed. Significant EVS coupling was observed with steering rate over approximately 0.5-8 Hz and roll rate over 0.5-4 Hz. Vestibular coupling with muscle activity was generally low and muscle-specific, with coherence observed over 0.5-3 Hz for flexor digitorum, 0.5-6 Hz for biceps brachii and erector spinae and 0.5-8 Hz for triceps brachii, deltoid anterior and deltoid posterior. Muscle response gain decreased significantly with increasing cycling speed for all investigated muscles, while coherence was largely maintained across speeds. Together, these findings demonstrate that vestibular perturbations systematically influence cycling balance through steer and roll kinematics and coordinated upper-body muscle responses. Greater steering and muscular response gains at lower cycling speeds, despite largely preserved coherence, suggest that vestibular input remains coupled to balance control across speeds while the magnitude of the resulting motor responses is modulated according to the dynamic stabilization demands of the bicycle-rider system. ... Read more

Cycling accidents are rising, and most one-sided crashes stem from a loss of control whose cause is rarely understood. The kinematics that would explain such events, such as roll, steering and crank angles in the moments before a crash, are not captured by the, sometimes, available video footage of the crash. This thesis asks whether these kinematics can instead be reconstructed from a single, static, monocular RGB camera.

The core obstacle is data. Unlike human pose estimation, there is no large public dataset of real bicycle motion with three-dimensional ground truth. The work therefore contributes a synthetic data generator that combines MuJoCo rigid-body simulation with photorealistic Blender rendering to produce physically plausible bicycle motion with exact two- and three-dimensional keypoint annotations. Second, it introduces 3D-BiKE, a modular pipeline tool that detects the bicycle with RF-DETR, estimates two-dimensional keypoints with RTMPose, and lifts them to root-relative three-dimensional keypoints with PoseMamba, a neural state-space model, from which roll, steering, and crank angle are extracted. Third, it assesses whether a state-space lifter designed for articulated human motion transfers to the rigid dynamics of a bicycle.

Trained only on synthetic data, the pipeline reaches a mean per-joint position error of 17.6 mm on validation data, better than human pose estimation methods trained on far larger real datasets. The derived roll, steering, and crank angles are recovered to 1.4°, 3.0°, and 5.1° root-mean-square error, substantially improving over the only comparable prior bicycle estimator. These results, though using only synthetic data, establish monocular keypoint lifting trained as a credible foundation for video-based cycling safety analysis, with validation on real footage as the essential next step. ... Read more

Cycling accidents among older adults caused by a loss of balance are a safety concern, particularly at low speeds where bicycles are inherently less stable. A Balance Assist system, which provide additional steering torque to support the rider, have been proposed as a way to improve stability and confidence for elderly cyclists. However, it remains unclear how different levels of assistance affect riding behavior and perceived comfort, and whether riders adapt to the system with repeated use.

This thesis investigated how different balance assist levels (High, Low, None) affect motion variability and perceived comfort, and whether repeated use across three trials leads to familiarization, reflected in reduced motion variability and increased comfort. Thirty-one older participants rode an instrumented TU Delft Balance Assist Bicycle, a bicycle developed in collaboration with Royal Dutch Gazelle and Bosch eBike Systems, without prior knowledge of the assistance system. They rode around a fixed course under one of the three balance assistance levels. Steering and roll behavior were recorded using sensors on the bike, and perceived comfort was measured using a questionnaire administered after each trial.

The results showed that balance assistance levels influenced riding behavior in a context-dependent way rather than through a general, uniform effect. In particular, higher balance assistance levels were associated with increased roll variability specifically in curved sections, while steering variability was primarily driven by riding speed and segment type rather than by the assistance level itself. Perceived comfort improved significantly across trials for all gain conditions, indicating that riders became more familiar with the bicycle over time regardless of the level of assistance received. However, this subjective improvement was not accompanied by significant changes in objective riding behavior, suggesting that familiarization occurred at the level of perceived comfort without corresponding measurable changes in motion variability.

These findings suggest that balance assist systems do not simply reduce motion variability across the board, but instead interact with specific riding contexts such as cornering in ways that require further investigation. The observed dissociation between subjective and objective adaptation highlights the importance of considering both perceived and measured outcomes when evaluating assistive cycling technologies for older adults. Future research involving longer-term use, system activation verification, and within-subject designs is recommended to better understand how balance assistance can be optimized to support safe and comfortable cycling in this population. ... Read more

Road-induced vibrations dissipate energy through the bicycle structure and rider’s body beyond what classical rolling resistance predicts. To measure this effect a laboratory apparatus is used that reproduces realistic road surface characteristics. This thesis presents the design, construction, and validation of such an apparatus and characterises the powerloss on a 3D-printed klinker road surface across six speeds (5–30 km/h) and five tire pressures (3.5–5.5 bar).
The setup measures the total resistive force acting on an athlete and racing bicycle using two loadcells: a main loadcell and an interface loadcell that captures the longitudinal force exerted by a two-bar linkage stabilization mechanism, isolated from the vertical load by a parallelogram flexure. Multiplying this force by the belt speed gives the powerloss. The force is measured with an expanded uncertainty of ±0.41 N at the 95% confidence level, corresponding to a powerloss uncertainty of at most 3.4 W at 30 km/h.
Powerloss increases approximately linearly with speed for all pressures, ranging from about 30 W at 5 km/h to 234 W at 30 km/h. Much of this loss is not classical rolling resistance: at 30 km/h the classical term accounts for only about 82 W, so roughly 64% of the total powerloss is attributed to vibrational losses. Tire pressure has a smaller effect: depending on the speed setpoint, an optimal tire pressure can be identified, although at lower speeds the error bars overlap and no clear optimum can be determined. The near-linear speed dependence shows that although the vibrational losses are large, their speed-growing quadratic component remains weak on the relatively smooth klinker surface (BRI ≈ 40). A stiffer underlayer and a rougher surface are recommended before the apparatus can be used to quantify vibration transmission from the road surface through the bicycle to the rider’s body. ... Read more

Single-track vehicle simulators have been identified as a subset of vehicle simulators not yet able to simulate sustained turning at slow speeds, a feature needed for increased simulator reality. Curiously,
most single-track vehicle simulators are built in similar fashion where infinite sustained turning is not a real possibility, virtually ignoring that other design solutions may exist. To this end, two new simulator concepts are designed that allow for infinite sustained turning, and these concepts are judged based on a list of fifteen design criteria. The resulting final concept is modeled using multibody dynamics, and simulating this model in multiple control scenarios shows that the concept is stabilizable while tracking a natural bicycle’s characteristic yaw rate to give the simulator the feel of a bicycle. Subsequently, the concept is further designed using CAD software and build, where the focus lay on the design being configurable, inexpensive, safe, free of backlash and sufficiently sturdy. As such, an inexpensive simulator was built using only a single actuator. The simulator has natural balancing and can do
sustained turning motions, whilst only taking steering as input. Preliminary testing revealed that humans can balance themselves on this simulator and that the simulator currently feels like balancing a bicycle at near zero velocity. Testing the simulator to figure out whether there was an optimal configuration between 3 variable parameters (physical size of the simulator, throttle curve and output configuration) revealed that positive throttle curves are necessary for easy simulator balancing. Of the 41 accepted trials out of 42 total trials, 0% of trials with negative throttle curves succeeded. Furthermore, the torque controlled output configuration had a 57.1% success rate while the PID speed controlled output configuration only boasts a 10% success rate. This difference may be due to an 8% input dead band imposed by the VESC, which removes all steering sensitivity in the ±7.5° area around the neutral steering position. This small area is believed to be especially important to the PID speed controlled output configuration. ... Read more

Anticipation is a vital component of cyclist behavior. Empirical evidence consistently suggests that cyclists do not merely react to immediate stimuli, but also actively predict future developments and adjust their motion accordingly. And such anticipative behavior varies across interaction scenarios. However, existing model-based studies provide only limited representations of how cyclists anticipate and how this anticipation varies, often relying on heuristic perception ranges, if-then decision rules, constant preview distance or fixed prediction horizons. Furthermore, anticipation-related parameters are rarely calibrated independently using real traffic data.

To address these gaps, this thesis proposes an adaptive-horizon Model Predictive Control (MPC) framework to explicitly represent cyclists' anticipative behavior. The proposed framework builds upon an existing MPC formulation for cyclist interaction and extends it by introducing an adaptive prediction horizon that varies with the inter-cyclist distance. We formulate the prediction horizon as a bounded and distance-dependent variable that can continuously adapt.

We calibrate the model parameters using two-cyclist overtaking interactions extracted from the TUMDOT dataset. The objective is to optimize MPC parameters to match the measured trajectory of the overtaking cyclist via a two-step Bayesian Optimization (BO) procedure. To disentangle the effects of the horizon-varying parameters from those of the remaining model parameters (background parameters), we implement the calibration sequentially. In step 1, we assume constant-horizon MPC and jointly calibrate the background parameters across all cyclist pairs within a given group. In step 2, the optimal background parameters are fixed, and the horizon-varying parameters are calibrated individually for each cyclist pair. We repeat both calibration steps under 6 BO configurations to reduce sensitivity to BO settings and to mitigate convergence to suboptimal local solutions.

The optimal constant-horizon model calibrated in step 1 achieves an average trajectory error of 0.3691 m relative to the measured trajectories, whereas the varying-horizon model calibrated in step 2 reduces this error to 0.1682 m, corresponding to an improvement of approximately 54.43%. In the correlation analysis of calibrated parameters, the practically used minimum prediction horizon is strongly correlated with its theoretical lower bound, indicating that this bound is often reached in practice. Also, the minimum horizon in all samples occurs at the overtaking point.

Furthermore, we observe moderate correlation between the calibrated horizon-related parameters and certain trajectory features. Higher speed is associated with smaller prediction horizon, indicating a shift toward more reactive behavior when cyclists move faster. And larger minimum inter-cyclist distance is associated with larger prediction horizon, suggesting that stronger anticipation may be associated with maintaining safer spatial margins during overtaking. ... Read more

This thesis investigates whether traffic patterns captured by a cyclist's camera coincide with moments of rider-reported workload increases, and whether a simple, scalable pipeline using a single forward-facing camera can extract useful signals for workload modelling. We developed an end-to-end system that detects and tracks nearby road users in cyclist point-of-view video, computing traffic features from 984 temporal windows across three urban rides in the city of Delft.

Using a compact set of eight traffic features and logistic regression modelling, we evaluated performance through three complementary approaches. Individual feature analysis revealed consistent directional differences between low and high workload episodes, with six of eight features maintaining the same directional relationships in multivariate analysis. Ranking performance exceeded baseline expectations, achieving within-route ROC-AUC of 0.570 (0.070 above chance) and PR-AUC of 0.450 (0.080 above baseline), meeting established criteria for small effect sizes. Cross-route generalization proved more challenging, with performance dropping to ROC-AUC of 0.517 and PR-AUC of 0.401, though consistently exceeding route-specific baselines across all test routes.

Binary classification demonstrated above-baseline performance, with within-route F1 scores of 0.478 representing a 0.108 improvement over our baseline expectations. Threshold optimization increased F1 to 0.517 by improving recall from 55.0\% to 78.3\%, though cross-route binary performance remained more limited. The traffic patterns associated with high workload episodes point toward defensive cycling scenarios characterized by increased object density, complex motion patterns, and reduced cyclist velocity in dense urban environments.

These findings establish that camera-derived traffic metrics contain detectable workload-related information while highlighting significant constraints for practical deployment. The modest magnitude of detected relationships and limited cross-route transferability indicate that while the approach demonstrates feasibility, substantial development is needed before robust real-world application. This work provides quantitative evidence for traffic-workload relationships in cycling and establishes methodological foundations for future research in automated cycling safety monitoring. ... Read more

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. ... Read more

As cities increasingly transition to car-free and low-emission zones, cargo bicycles have become a popular alternative for both companies and families. They allow for the efficient transport of goods and children while navigating busy urban areas without emissions. However, single-track cargo bicycles still have some limitations. Stability issues, especially at low and zero speeds, and the effort required to lift a heavy cargo bicycle onto its kickstand, reduce usability and safety. These problems are amplified by uneven load distribution and heavy cargo.

A prototype for an active kickstand stabilization mechanism was developed. A model was made where a simple spring-damper feedback system was placed on the roll degree of freedom of the cargo bicycle. After formulating design criteria, a design was created and manufactured, and the system was assembled on a cargo bicycle. The prototype was then tested to validate the design specifications.

The results show a measurable improvement in stability, especially when cycling off from a standstill. The system also stabilized the bicycle with a cargo of 50 kilograms on the luggage rack in addition to an 80-kilogram rider. These results show that an active kickstand stabilization mechanism is suitable for everyday use on cargo bicycles and can increase the safety and usability of the cargo bicycle. ... Read more

Bicycles have seen relatively few safety features added since their invention, especially in comparison to the numerous advancements in automotive safety. This thesis investigates the use of simulation to model bicycle dynamics, progressing from a no-slip model to a more complex model incorporating slip, and finally to a model that simulates the interaction of a bicycle with an unexpected patch of ice. Tire slip on ice was estimated using car tire slip data combined with bicycle measurements taken on dry asphalt. The study evaluates the effectiveness of two control strategies, Proportional-Derivative (PD) control, and Linear Quadratic Regulator (LQR) control, through these simulations. The PD controller was tested with gain values ranging from 0 to 20 at regular intervals, while the LQR controller’s weightings were selected through a process of trial and error. The analysis focuses on how variations in PD gains and LQR weightings affect the bicycle’s stability and ability to traverse icy terrain. Results indicate that the LQR controller consistently stabilized the bicycle at an 8° lean angle across all tested speeds and maintained stability at a 12.5° lean angle at higher speeds when crossing ice. Ultimately, the selected LQR controller used diagonal values of 50 and 5 in the state weighting matrix Q, and a value of 1 in the control weighting matrix R.These findings suggest that advanced control strategies, particularly LQR, can enhance bicycle stability in adverse conditions ... Read more

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. ... Read more

Bicycles are among the simplest and at the same time most fascinating vehicles. Despite their apparently basic design, their stability, dynamics, performance, and safety are still largely unknown. In a world where cities suffer from critical air pollution levels and are often congested by car traffic, bicycles can represent a cost-effective and ready solution to address part of the problem. However, the increasing use of bicycles enlarges the risk of bicycle-related accidents and injuries. Many of these accidents involve falls, highlighting the need of proper studies on bicycle-rider dynamics. While research has been conducted on the effect of bicycle parameters on the handling quality properties, we still do not know enough about the specific role of tyres on bicycle-rider dynamics. To better simulate what happens in the real world, existing bicycle mathematical models need to be integrated with robust tyre models that include side slip, as they can change both the lateral and longitudinal dynamics. Actually, the use of nonlinear tyre models allows taking into account the lateral force and self-aligning torque saturation. This is needed for instance to model the wobble as a non-divergent vibration modes, contrarily to what happens when using a simple linear tyre model. These features relate directly to predict critical situations for which we need a deep knowledge of bicycle tyres, thus proper test-rigs to measure their mechanical characteristics... ... Read more

In this research a dynamic time and energy model is constructed to simulate free flow utility cycling. Using a Monte Carlo simulation and distributions of rider and bicycle characteristics a population of cyclists is modelled to find the difference in time and energy expenditure on six different bicycle routes. The riders use realistic like power inputs and braking behaviour based on the rider and the route characteristics. While travel times are almost always lower for all cyclists when comparing routes, there are cyclists for who a shorter route distance does not equal a lower energy expenditure. For simpler routes (less elevation difference, traffic lights and shorter distance) the standard deviation of both time and energy decreases, showing that slow cyclist have a relative higher gain. For e-bike users there is even almost no difference in energy expenditure between the six evaluated routes that have varying elevation, traffic signals and routing, while travel times show a similar trend as for regular bicycles. With this model difference in routes can be quantified in matters of time and energy expenditure for a population of cyclists giving an objective picture of the differences between routes, which can be a useful tool for city planning and evaluation of bicycle infrastructure. ... Read more

Model matching control allows for a virtual change in the vehicle's physical parameters, removing the need to alter the system physically. These vehicles are used as a versatile test bed for safe and repeatable experiments, and allow for safer vehicle operation by virtually keeping the driving behaviour constant in the face of physical changes such as wet roads. Although already applied on planes and cars, it has not been applied to bicycles. The proposed control method accomplishes model matching control by equating the state derivative of the physical system to that of the the virtual system. Theoretical analysis shows model matching control is applicable to bicycle dynamics, even if only the steer torque is available for control. Via experiments, I identify the bicycle's weave mode eigenvalue at different speeds, and the Bode gain between steer torque to lean rate and to fork angle for frequencies between one and three Hz. The experimentally identified values follow the theoretically predicted values closely but do not overlap perfectly. The effect of artifacts during the experiment and unmodelled dynamics are analysed with the use of simulations and are able to explain these perceived differences. From the results, this research concludes that model matching control is applicable to bicycles and is able to virtually change the bicycle's physical parameters. Future work will look into making the controller more robust to modelling errors. ... Read more

This report describes the development of a Python based tool for 2D structural analysis in civil engineering, using the SymPy library (Meurer A, 2017) and applying the Macaulay method (Macaulay, 1919) to enable symbolic computation for structural analysis. The tool serves as a proof of concept demonstrating the feasibility of symbolic computation in analyzing internal forces. Central to the project is the extension of existing SymPy modules, specifically the Beam module, to support 2D analysis by introducing new modules: "Column" and "Structure2D." The existing "Beam" module in SymPy, which computes vertical load effects, was used to handle shear forces and bending moments by creating a load equation for the vertical direction. To handle normal forces a "Column" module was proposed (although not implemented in this iteration), designed to compute horizontal load effects based on the material’s elasticity modulus, length, and cross-sectional area. The "Structure2D" module integrates both vertical and horizontal load computations by transforming a 2D structure into two corresponding 1D representations. This method enables separate computations for each direction, converting 2D problems into simpler 1D computations while preserving load and support behavior. The tool’s workflow involves users defining structure components (members, loads, supports) on a 2D grid. Once defined, the structure is "unwrapped" to map each member along a continuous line. Each load is split into its horizontal and vertical components and applied to the corresponding axis. Through examples, the report showcases the tool’s ability to calculate shear forces, bending moments, and reaction loads. Although the project successfully demonstrates the potential of symbolic computation for 2D structural analysis, several limitations exist. The tool currently lacks a functional Column module, restricting it to single horizontal reaction loads and limiting its ability to solve for horizontal deflections and axial forces fully. Additionally, the structure must be non-branching and linear, excluding complex configurations like truss networks. Material properties must be uniform across members, a constraint that simplifies the model but limits its real-world applicability. Addressing these limitations would involve developing a more advanced unwrapping algorithm for intersecting members and introducing a complete Column module. In summary, this prototype validates the possibility of a symbolic 2D structural analysis tool, providing civil engineering students and educators with an analytical framework that combines Macaulay’s method and modern symbolic computation. This tool lays the groundwork for future developments that could expand its functionality, to move away from proof of concept to a fully featured open-source product ... Read more

Bicycles have been studied extensively over the past 200 years, with mathematical models providing valuable insights into various aspects of bicycle dynamics and rider control. However, the lack of a common framework for creating and sharing bicycle-rider models hinders the development of advanced models, research reproducibility, and dissemination. This thesis addresses this gap by introducing BRiM: an open-source modular and extensible framework for creating Bicycle-Rider Models.

The modular setup of BRiM relies on a systematic approach to define a model and form the analytical equations of motion. For the involved analytical computations BRiM utilizes SymPy, a Computer Algebra System. The systematic approach consists of four stages. The first stage defines the objects in the system, such as symbols and bodies. Secondly, the kinematic relationships between the objects, such as angular velocities between reference frames, are established. The third and the fourth stages, which are order-independent, specify the loads and constraints acting upon the system. The decoupling BRiM required to achieve modularity is enabled through this systematic approach, because computations within a stage are mostly order-independent.

The core of BRiM employs the systematic approach within a unified framework for modeling mechanical systems in general. It describes a model using a tree representation, in which a model is defined as an aggregation of smaller submodels. The relationships between submodels are established by parent models, using interchangeable connections to accommodate complex relations, such as tyre models between the ground and a wheel. This application of submodels enables swapping and adding submodels, making the overarching model both modular and extensible. Actuation within BRiM can either be specified by attaching prespecified groups of loads to models and connections, or by utilizing the interface provided by the mechanics module in SymPy, which offers the flexibility to even manipulate equations in detail.

BRiM applies this generalized framework to create modular bicycle-rider models. Both a stationary bicycle and a modular bicycle based on Moore's convention of the Carvallo-Whipple bicycle have been constructed. These bicycle models are extensible to bicycle-rider models by including an upper and/or lower body. Within the rider models each joint can be actuated by a linear torsional spring-damper. BRiM integrates parametrization of models, which provides mappings between symbolic quantities used in equations and experimentally determined values, using the existing open-source BicycleParameters library. Additionally, SymMePlot, a visualization package for symbolically defined mechanical systems, has been developed and integrated within BRiM to visualize the created bicycle-rider models.

The effectiveness of BRiM is demonstrated through optimization and simulation tasks. Firstly, a real-time forward simulation of a torque-driven upper body bicycle-rider is performed. Secondly, an optimization problem is solved, involving the tracking of a rolling disc along a sinusoidal trajectory while minimizing the control torques. These demonstrations highlight the seamless integration of BRiM with other scientific tools and BRiM's potential for practical applications.

In conclusion, BRiM fills the gap in bicycle dynamics research by providing a modular and extensible framework for creating and sharing bicycle-rider models. Its systematic approach, unified framework, and integration capabilities enable efficient model development, research reproducibility, and further advancement in bicycle research.
... Read more

This study investigated the relationship between Equivalent Fall Height (EFH) and peak Ground Reaction Force (GRF) in freestyle ski jumping. An experiment was conducted in which a semi-professional freestyle skier was instrumented with Inertial Measurement Units (IMUs) and a position sensor. The skier performed jumps off ‘kickers’ in snow parks in four different ski resorts in Northern Italy. The gathered sensor data was utilized to calculate the EFH and GRF of every jump. The data reveals convincing evidence that a positive correlation between these two parameters exists. However, the coefficient of determination, R2, indicates only a moderate correlation. The results suggest that alternating landing strategies might have affected the measured GRF as well. This indicates that a higher correlation between EFH and GRF might exist than shown in this research. Further research should focus on how individual parameters, such as EFH and landing strategy, affect the GRF. This should develop a more comprehensive understanding of how to enhance the safety and enjoyment of freestyle ski jumps for a broader audience. ... Read more

The second most used way of transport in the Netherlands is the bicycle. There is a great variety of city bicycles in which each person has their preference. One of the most critical aspects of personal preference is the cycling experience or feeling, better defined as Handling Qualities. Improving the Handling Qualities can lead to more comfort and safety in cycling. The goal is to determine what could improve Handling Qualities which will help within the design stage of bicycles. Handling Qualities are different for each bicycle and depend on many various aspects of the bicycle's geometry. In addition, there are differences in riders' posture and stature that have an influence on the bicycle's control.
To find the most critical parameters that influence the handling qualities, three different bicycles are analysed. Furthermore, three different postures or sitting positions (forward lean) are compared to find the influence of the rider on the Handling Qualities of the total human-bicycle system. In addition, three different statures combined with the right side of the bicycle are compared. A database for human measurements is used to calculate the rigid body of the rider based on an anthropometric model \cite{Moore2009}. The bicycle parameters are gathered from Solidworks bicycle models, given by Gazelle.
A dynamical model is created based on the Whipple bicycle model \cite{Meijaard2007b} with an arm model extension \cite{Schwab2012}. (Steering is the most essential control input).
The eigenvalues are calculated to analyse the stability of each combination. Furthermore, the Handling Quality Metric (HQM) \cite{Hess2012} is calculated to compare Handling Qualities.
As a result of using an arm model, there is no self-stability within any of the rider-bicycle systems. The system can still keep the bicycle upright due to the human arm control of the bicycle. An analysis of eigenvalues shows that the dynamic behaviour of each combination is greatly influenced by the arm model. The different sitting positions and statures also gave a great variety of dynamic behaviour.

The HQM values are greatly influenced by the added arm model, some rider bicycle combinations gave extremely high values. However within high values, there were varying results; for every combination, the HQM values were best for the most upright sitting positions. Furthermore, the best HQM values found are for the tallest person and the worst for a very small person.

Both dynamic behaviour and HQM values are greatly influenced by the added arm model. However, the differences between postures and statures have become clear within this model. An increase in stature has a positive influence on the HQM values as well as a more upright position.
The model can be used in the design stage of the bicycle to predict handling qualities according to the HQM standard. Influential parameters for the desired handling qualities can be found. Different sitting positions and people can be evaluated. Therefore, bicycles can be designed for specific target groups. ... Read more

Most of the injuries sustained by cyclists are due to accidents that do not involve other road-users. This type of accident is known as a single-bicycle accident. A bicycle that is more stable may help to reduce the number of single-bicycle accidents. A balance-assist bicycle is developed at the TU Delft in collaboration with Royal Dutch Gazelle and Bosch eBike Systems. This bicycle has a motor that applies a torque about the steering axis. The motor automatically steers into the direction of the fall - the primary method of controlling a bicycle. The goal of the balance-assist bicycle is reducing the amount of single-bicycle accidents by increasing the stability of the bicycle.

The first research goal of this thesis is to characterize the dynamics of the balance-assist bicycle. Experiments are performed at multiple speeds and controller gains in which the riderless balance-assist bicycle is perturbed and the consequent roll rate is measured. A function describing a decaying oscillation is fitted to this data to retrieve the eigenvalues. A Whipple-Carvallo parameter set is created that approximates the measured dynamics of the bicycle. The results show that the balance-assist controller is not able to stabilize the bicycle when the mass and inertia of a rigid rider is added. However, the controller does damp the weave mode substantially. This likely makes it easier for an active rider to control the bicycle.

The second research goal is to create a model based on experimental data that predicts what perturbations to the rider-bicycle system result in a fall, and what perturbations do not result in a fall. An experimental setup is created that can apply perturbations to the bicycle in a safe environment. The experimental set up consists of four motors that pull on ropes attached to load cells at the end of the bicycle's handlebars. Experiments take place on a treadmill surrounded by padding. Participants wear a safety harness suspended from the ceiling. An operator controlling the treadmill stops the treadmill when a participant falls. The four motors are able to apply perturbations of different magnitudes and direction to the handlebars of the bicycle. The goal of the experiments is to apply perturbations for which a participant falls approximately 50\% of the
time. The controller applying the perturbations and internal sensors of the balance-assist bicycle collect data.

A Bayesian and a frequentist multilevel logistic regression models are developed to predict the probability that a perturbation results in a fall. Results show that there are two predictor variables that have a significant influence on fall probability. Firstly and most importantly, an increase in angular impulse increases the probability that the cyclist falls. Secondly, the fall probability decreases if a participants has experienced more perturbations.

The created model is used to predict the effect that the balance-assist controller has on the fall probability. This shows that the controller could substantially reduce this probability. Moreover, this thesis shows that the balance-assist bicycle substantially dampens the weave mode of the balance-assist bicycle when the mass and inertia of a rigid rider are added. This likely makes it easier for a rider to recover from a perturbation. So, results from this thesis show that, in theory, the balance-assist bicycle could reduce fall probability. The experimental set up, protocol and data analysis scripts developed in this thesis provide an excellent starting point to validate the effectiveness of the balance-assist bicycle in future work.
... Read more