Loss of control inflight is the most common cause of fatal accidents in aviation. Aerodynamic stall models are utilized in pilot training to enhance safety and prevent accidents. This research presents an advanced longitudinal stall model for the Cessna Citation II, achieved through innovations in modeling methodologies and experimental design. By introducing dynamic stall maneuvers with step inputs, the study mitigated $\tau_1$ and $\tau_2$ parameter correlation, enabling more reliable parameter identification. A separable nonlinear least squares method significantly reduced computational time for nonlinear stall model estimation, decreasing it from hours to seconds. This approach revealed two minimally correlated flow separation states, offering deeper insights into wing flow characteristics and improving model accuracy. The lift model was refined to incorporate pitch rate and elevator deflection effects, while the drag model was enhanced with a lift-induced drag component. Additionally, a center of pressure model was derived from pitching moment data, advancing the understanding of stability during stall. A novel structure for characterizing degraded elevator control effectiveness was also developed. These advancements resulted in substantial performance improvements, with mean squared errors for lift, drag, and pitch moment coefficients reduced by 32\%, 29\%, and 27\%, respectively. The models also demonstrated greater consistency across diverse maneuvers, evidenced by reduced variability in $R^2$ values. This work contributes to more accurate stall modeling, enhancing both aerodynamic understanding and aviation safety.

One of the most widely applied identification methods for stall modelling from flight test data is based on Kirchoff’s method of flow separation. This approach has not lead to a satisfactory aerodynamic pitching moment model. The introduction of the so-called X-variable, representing the point of flow separation on the wing, interferes with identification of a pitch damping term, that is needed for dynamic stability. Moreover, flow separation is only a small contributor to the pitching moment, leading to a lack of physical interpretability. In general, Kirchoff methods lead to models incompatible with the nominal flight envelope. This paper presents a nonlinear unsteady model of the pitching moment using lag states of the angle of attack measurements, identified from flight test data collected with a Cessna Citation II. The model is formulated in terms of well-known stability derivatives and is a one-on-one extension of the nominal envelope model. Model regressors are selected from a large pool of candidates using Multivariate Orthogonal Function Modeling. The candidate pool is based on a newly formulated mathematical model, such that each model contribution has a clear physical interpretation. This has lead to a 𝐶𝑚𝛼 contribution depending on various lag states of the angle of attack, and pitch and downwash lag damping contributions as univariate splines. The model has good predictive abilities and can report a reduction of 55.9% in validation MSE compared to a Kirchoff based pitching moment model by van Ingen et al..

The primary objective of this thesis is to develop a new model that accurately predicts vertical mo- tion sickness by closely mirroring the neural dynamics and statistical properties of the otolith pathway. Unlike existing models, our approach incorporates elements such as noise, nonlinearities, and realistic sensory data, enhancing the accuracy and biological relevance of motion sickness predictions. This re- search aims to create a biologically plausible approach that derives a proxy metric for Motion Sickness Incidence (MSI). This metric will naturally emerge from sensory conflict, providing a more accurate prediction of vertical motion sickness and contributing to the development of systems designed to mit- igate its effects. This has become increasingly critical with the growing prevalence of self-driving cars and virtual reality systems, where effectively addressing motion sickness is of great value. To achieve this goal, we developed a novel model from the ground up, incorporating detailed dynam- ics and statistical properties of otolith hair cells and afferents in the inner ear. Three spiking models were developed based on Leaky-Integrate-and-Fire, Hodgkin-Huxley, and Generalized Linear Model frameworks. Each model presents unique advantages in capturing neural behavior. Our findings indi- cate that while all models offer valuable insights, the Leaky-Integrate-and-Fire model most effectively captures the dynamics, statistics, and temporal characteristics of otolith afferents. The spiking models were used in a scheme that improves upon existing models in three ways. Firstly, and most fundamentally, we added noise and probability elements to our modeling. This makes the models more realistic by reflecting the natural randomness and variations in human behavior and per- ception. By more closely mimicking real-world conditions, our models better represent how humans perceive and respond to stimuli, making the simulations more accurate overall. Additionally, we incorporate nonlinearities to accurately represent sensory saturation and have imple- mented a realistic organ definition, including hair cell tuning. The model relies on actual sensory data rather than simply aligning with subjective experimental results, allowing for a more accurate and bio- logically relevant simulation of sensory processing. Lastly, the spiking models are integrated within a framework that utilizes Bayesian inference to esti- mate vertical acceleration, taking into account the inherent noise in the brain and aligning with modern neuroscience theories. By bridging the Free Energy Principle with the Sensory Conflict Theory, our model provides a more robust framework for understanding and predicting motion sickness. The model gives results that are representative in terms of human motion perception sensitivity, while also being in line with current knowledge about frequency and amplitude dependence of Motion Sickness Incidence. The resulting model gives us insights into how otolith dynamics influence motion sickness and is a proposal on how the brain resolves estimating vertical motion perception. Using different model configurations we were able to create a model that estimates motion accurately and propose a proxy metric for motion sickness incidence. This model is created in a modular way and can be extended on the basis of new objectives or new research findings. Additionally, with further enhancements, it holds potential applications in validating otolith theories on statistics and information transmission characteristics and even in the development of prostheses.

The human operator's manual control behaviour under time-invariant conditions has been successfully modelled. However, significant gaps remain in understanding and modelling adaptive manual control behaviour under time-varying conditions. One such less-understood aspect is the pilot's ability to detect changes in time-varying controlled element dynamics. This study aims to develop a mathematical model that investigates open-loop stability as a criterion in compensatory and pursuit tracking tasks to model the pilot's detection of change in controlled element dynamics across different transition rates, particularly for transitions from stable to less-stable vehicle dynamics. The model operates under the assumption that trained human operators track statistical properties of tracking task signals within periods of compromised open-loop stability to trigger their detection of change in dynamics. The model identifies regions of reduced stability and simulates the tracking task signals through time-varying computer simulations. Subsequently, human-in-the-loop experiments are conducted to validate the model. The validated model demonstrates a combined accuracy of $88.54\%$ for the compensatory task and $80.62\%$ for the pursuit task. Notably, in the experiments, the error signal consistently outperforms the error rate signal across all transition rates, diverging from the results of the computer simulations. Overall, the proposed model's ability to successfully predict pilot detection across a spectrum of transition rates marks an advancement towards developing more human-like automation.

In Haptic Shared Control (HSC), human-like reference generators and adaptive strategies have shown promising potential for minimizing human-machine conflicts, though these advances have thus far been limited to simple control tasks. This paper extends the application of low-conflict HSC to a more realistic driving task. An Adaptive HSC (A-HSC) design for constant-velocity steering under changing visibility is proposed, where the A-HSC dynamically adjusts its steering support to align with the human driver's behaviour. In a human-in-the-loop simulator experiment with 16 subjects, the proposed A-HSC adapted successfully to the drivers' steering behaviour, converging to an average look-ahead time of 0.46s in low visibility and 1.01s in high visibility. When visibility decreased during the task, the driver's trust and control authority influenced the adaptation, with half the drivers complying with the A-HSC support allowing it to retain high-visibility settings. The A-HSC outperformed the fixed low-visibility HSC system in reducing driver control effort, minimizing conflicts and improving subjective acceptance. However, it ranked lower than the fixed high-visibility HSC overall. To further enhance A-HSC adaptability and driver acceptance, future research should investigate how varying the authority balance between the HSC and the driver affects adaptation, and explore more intuitive HSC structures, potentially based on the driver's visual aim point.

The increasing demand and complexity of air traffic management (ATM) systems necessitate significant advancements in automation to ensure safety and efficiency. Artificial intelligence (AI) and machine learning (ML) are emerging as promising solutions to manage this growing complexity, offering enhanced decision-making and predictive capabilities. However, the effectiveness of ML models in ATM heavily relies on the availability of extensive, high-quality data. In many cases, such data is scarce or incomplete, which presents a major challenge for training robust models. Synthetic data generation (SDG) represents a viable solution to address this data scarcity, enabling the creation of realistic, diverse datasets that can effectively fill data gaps and unlock the ML value proposition. The Terminal Maneuvering Area (TMA) is a crucial segment of airspace where aircraft perform complex maneuvers including rare go-arounds, and holding patterns. The TMA is characterized by high traffic density and diverse trajectory types, necessitating precise, granular data to model these dynamic scenarios accurately. The main research objective of this thesis project was to investigate the applicability of Generative Adversarial Networks (GANs) in generating synthetic 4-dimensional aircraft landing trajectories capable of capturing traffic patterns in this airspace. An unconditional recurrent-based network (TimeGAN) was applied and complemented with an investigation into the merit of conditioning to reap the rewards of transfer learning between classes. The resulting synthetic trajectories were evaluated in terms of data diversity, fidelity and usefulness. The main challenge identified during this research was the imbalance in data classes, which affected the models' ability to accurately capture data patterns, particularly in less frequent scenarios. Generating synthetic data based on separate groupings showed promise in addressing these imbalances, although this approach was sensitive to the number of clusters chosen. Smoothing filters frequently proved necessary to ensure realistic landings were devoid of unwanted noise. Overall, this work proves the capability of TimeGAN in generating diverse, realistic trajectories that are difficult to differentiate from real historical data.

Relocated haptic feedback from the fingertip to the proximal phalanx can alter the perception of physical interactions by simultaneously displaying relocated multimodal tactile cues. However, how these relocated tactile cues alter the vibrotactile sensitivity of the fingertip remains unclear. This two-site stimulation study employs a customdesigned multimodal haptic ring for the proximal phalanx to evaluate the effect of relocated cold, hot, and pressure stimuli on the vibrotactile sensitivity of the fingertip. Our results show no significant difference between the vibratory detection thresholds of six multimodal tactile conditions. In contrast to single-site multimodal stimulation, where vibrotactile sensitivity is significantly altered by pressure and thermal stimuli, these results imply the feasibility of applying thermal and pressure stimuli without significantly altering fingertip sensitivity. Relocation of tactile stimuli keeps the vibrotactile sensitivity of the fingertip intact and opens up the possibility of altering tactile perception while interacting with the physical environment, making this technology feasible for mixed reality applications.

Commercially available Lane Keeping Assist systems fail to consider the driver's intentions since they mainly focus on minimising path tracking errors, resulting in conflicts between humans and automation. This often leads to users being unsatisfactory and turning off the assist, as a result diminishing the advantages such as reduced workload and increased road safety. Considering a driver model in the assist helps increase user acceptance. Therefore, we propose a torque-based hybrid controller for a human-centric haptic shared Lane Keeping Assist, pairing a data-driven driver model with a model-based controller to foster the collaboration between the driver and assist. First, the driver's non-linear steering wheel torque behaviour is modelled and predicted using a Bidirectional Long Short-Term Memory network with an accuracy ≥72.4% and a smoothness ≥0.85Nm/s over a 0.4s prediction horizon. Second, a Model Predictive Controller with a linear bicycle and steering model is developed, where it utilises the driver model's predictions as a time-varying reference. We developed three human-centric controllers for comparison and used a state-of-the-art commercial solution as the baseline controller. The experiments were performed in Toyota Motor Europe's fixed-base driving simulator, where 15 participants tested and evaluated the four controllers. The results show a 113.1% increase in collaborative ratio while maintaining a similar path tracking performance compared to the baseline.

The identification of time-varying, adaptive behaviour of a human operator in basic manual control tasks is undoubtedly under development since most methodologies only account for time-invariant systems. Previous authors have proved that estimation techniques based on ARX structures can generally identify the HO model parameters. Nonetheless, ARX methods present several problems, such as the persistent bias in estimates that may increase due to coupled noise and system models. Therefore, a novel identification technique based on Box Jenkins models is proposed to achieve a more adequate match between the estimator structure and the HO model. The identification process can be conducted offline by the Ordinary Least Squares and Prediction Error Method, or online, when Recursive Least Squares and Recursive PEM are employed, respectively, in ARX and BJ models. The BJ estimator has excellent potential as an identification tool due to its bias reduction capabilities, as clearly shown in batch-fitting, although non-linear optimization processes decrease its convergence speed by 500%. An RPEM algorithm with forgetting factor 𝝀 = 0.99609 and first-order remnant BJ structure is implemented and tested under Monte Carlo simulation and experimental data. Recursive BJ algorithms could help to achieve the ideal identification method by diminishing the Neuro-Muscular parameter bias in ARX.

Aerial manipulators, characterized by their ability to actively engage with the environment, are gaining popularity for their versatility in performing diverse tasks.
This research focuses on augmenting the capabilities of aerial manipulators through the integration of tactile feedback, specifically employing a compliant bio-inspired three-fingered manipulator equipped with tactile capacitive sensors on each finger. The manipulator is affixed to a drone, enabling tactile-guided navigation for precise object localization, subsequent grasping, and perching. Additionally, a grasp evaluator assesses grasp quality, allowing the system to adapt by suggesting alternative grasp locations after an initial attempt is unsuccessful. A comparative analysis between the system’s performance using tactile feedback and open-loop perching/grasping in perching scenarios demonstrates that the grasp evaluator improves the perching success rate by 55%-point and increases the allowable object uncertainty by 0.14 [m]. These findings highlight the efficacy of this approach in advancing aerial manipulator capabilities.

Navigating touchscreens in vehicles becomes increasingly challenging when subjected to external perturbations like air turbulence or bumpy roads. These perturbations can lead to a loss in task performance, reduced finger accuracy, and increased frustration among users. To address these issues, electrovibration has emerged as a promising technology to enhance touchscreen interactions by providing users with better feedback and maintaining touch stability even under challenging conditions. Before we investigate if electrovibration helps users in challenging conditions, first the effects of external perturbations on tactile perception by electrovibration must be found. To investigate the impact of external perturbations on electrovibration perception, we conducted psychophysical experiments with 18 participants interacting with an electrostatic display mounted on the cockpit of the SIMONA flight simulator. We measured participants’ absolute detection thresholds for electrovibration pulses generated using 100 Hz input voltage for durations of 0.2 and 0.5 seconds, simulating a ridge and a button or slider. The measurements were taken under no-turbulence conditions and two different turbulence conditions. Our results revealed that turbulence significantly affects vertical finger movement, average normal force, and the change in force applied to the screen by participants. This combination of factors, along with the 0.2-second pulse duration, makes electrovibration more difficult to perceive or even imperceptible. The 0.5-second electrovibration pulse was perceived better overall and remained unaffected by turbulence. This highlights the strong significance of pulse duration on the absolute threshold of electrovibration. Therefore, to counteract the negative effects of perturbations on perception, electrovibration should be employed for longer durations when perturbations are present.

When placing a Head-Mounted Display (HMD) of a Virtual Reality (VR) or Augmented Reality (AR) system in the cabin of a moving base simulator to provide a larger field-of-regard, the HMD's inertial sensors will interpret the motion of the moving base simulator as if provided by the user, thus distorting the HMD's cabin-fixed pose estimation.
In order to estimate a proper cabin-fixed HMD pose, this thesis proposes an Unscented Kalman Filter (UKF) that fuses the information from the simulator sensors, as well as the sensors of the VR/AR system, including its visual position sensor and IMU. For this, three sensor configurations are proposed and tested offline deterministically based on real data. Although an unexpected latency in the HMD's visual pose sensor prevented a full validation of the algorithms, both the configuration with an IMU attached to the motion base, and the configuration without an IMU but with second-order additive noise variables added to the UKF, resulted in a proper, tight innovation sequence, indicating accurate system state estimation. The third configuration, one using the novel approach of using setpoints, provided a tight, but out-of-bounds, innovation sequence indicative of high accuracy, but also diminished tuning efficacy. It is also sensitive to stability issues when estimating the motion system's dynamics when no movement is present on the moving base. The unaltered setpoint method is not recommended as a solution. However, a hybrid solution of the setpoint and second-order additive noise variables methods should be investigated in a follow-up study to combine stability with high accuracy. Another recommendation is to test all configurations interactively, i.e. online.

During human motor control, sensory information is integrated, and the influence of each cue depends on its variance. Sensory weighting within the proprioceptive system, specifically between force and position feedback, including its stiffness dependency, has been demonstrated for the shoulder and digits. Both studies encompass upper limb joints, but the impact of joint proximity on sensory integration remains unknown. In this study, we analyzed isolated vertical reaching movements of the wrist (palmar flexion), elbow (flexion), and shoulder (retroflexion) using a haptic robotic manipulator. We varied the stiffness constant of a virtual spring (15 N/rad, 50 N/rad, 80 N/rad) to assess the influence of environment stiffness on participants' ability to blindly reproduce a target force. A nonlinear spring was introduced to reveal the weighting strategy between force and position feedback. Differences in force and position when participants unknowingly operated in the nonlinear environment, compared to the linear environment, allowed us to calculate the weights of both feedback cues. Repeating the experiment for the three joints allowed us to assess the influence of joint location on the weighting strategy between force and position feedback. Ten participants performed a total of 720 reaches, covering three different stiffness conditions, three distinct joints, and ten separate runs, each consisting of eight reaches. We hypothesized that both environment stiffness and joint location have a significant impact on the weighting strategy. We expanded on existing findings that show a tendency toward up-weighting force feedback in the integration process as environment stiffness increases. Additionally, we examined whether force or position feedback is favored when the joint is more proximal. The hypothesis is supported by experimental evidence, as both stiffness and joint type revealed statistically significant effects on the weighting strategy (p $<$ 0.001, p = 0.020, respectively). As environment stiffness increased, a shift from higher position weighting factors to higher force weighting factors was observed for all three joints. When comparing our findings to the Maximum Likelihood Estimation model predictions, we observed that the more proximal joints exhibited higher force weighting factors. Additionally, normalizing the experimental environment stiffness conditions over the maximal voluntary force of each joint times the moment arm and over the joint stiffness showed a similar trend. More distal joints, on the other hand, favored higher position weighting factors.

It is impossible to imagine modern day interaction with technology without the use of touchscreens. It is a go-to interface to use for many applications, because of the high stimuli-response compatibility and adaptability of the graphical user interface. But the haptic feedback one would have with physical buttons and dials, is lost with the use of touchscreens. High potential to improve the interaction with high resolution haptic feedback is often ignored. In this paper, the use of a haptic pseudo-potential field rendering method on a friction modulated touchscreen is proposed. With this method, the user is assisted in moving towards a target by lowering the friction and impeded in moving away by increasing the friction coefficient. In a human factors experiment, this rendering method is compared to a position-based friction modulation method. Subjects are instructed to find a target path, based on the haptic feedback. The results show that the position-based rendering method has a higher hit-rate and lower movement times. This demonstrates that the pseudo-potential field method is difficult to perceive, however it is expected that advancements in rendering larger friction coefficient ranges or even active lateral force feedback will improve this rendering method.

Thermal feedback has been proven to enhance the user experience in human-machine interaction. However, state-of-the-art technology mainly focuses on static contact using either palm or fingertip, overlooking dynamic and multi-finger interactions. Underlying challenges include incompatible designs of the conventional interfaces for providing controllable salient thermal stimuli for such interactions and, thereby, lack of knowledge on human thermal perception for relevant conditions. Here we designed a new thermal display that can deliver distributed spatio temporal thermal patterns and investigated the influence of user exploration on the perception of these patterns. Twenty-three human participants interacted with the device using three exploration conditions (static-single finger, dynamic-single finger, and static-multi finger) and evaluated 15 temperature differences ranging from +1.5◦C to -7.5◦C. Our results showed that humans are significantly more sensitive to thermal stimuli when exploring via static single-finger contact than other tested conditions. Moreover, in the case of static-single finger interaction, we found larger thermal discrimination thresholds compared to the literature. Our findings offer new perspectives on providing salient and consistent thermal feedback for future tactile interfaces.

Research into Aerodynamic Stall Model Identification (ASMI) is an ongoing endeavor to create a model that achieves the highest fidelity, that aids in upset recovery training of pilots in Flight Simulator Training Devices (FSTD). Aircraft come in various configurations, one aircraft stall model cannot sufficiently provide a full understanding regarding these differences in configuration. Obtaining proper flight-test data is a time-consuming, and a very expensive exercise, especially when large commercial aircraft are concerned. The partnership with Fokker allows for a highly appreciated, and novel step in stall model identification.

A first step was set in the data preparation phase, where historic flight-tests are intended to be used towards research into aerodynamic stall model identification. This led to the creation of a fuel depended, highly accurate mass model, and an engine model based on OLS estimation techniques. Finalized by setting up Kalman filtering techniques for estimating the aircraft state.

Understanding human perception of haptic feedback is critical when designing and regulating these control systems. In recent years, experiments have been conducted to determine the Just-Noticeable Difference (JND) in mass-spring-damper dynamics using a hydraulic admittance display in the form of a side-stick. These experiments have resulted in a model of JNDs when interacting with linear second-order dynamics. In real-world applications, however, control force dynamics also commonly include nonlinearities like friction. This research extends the current understanding of JNDs in linear systems by including the nonlinear case, where friction is also present. Experiments were conducted to determine JNDs in friction when combined with second-order system dynamics. Initial results suggest that friction JND can be independent of linear system dynamics as long as its value compared to the linear system’s impedance is sufficiently large. This means, friction JND follows Weber’s law, also when it is combined with mass-spring-damper dynamics, unless the level of friction approaches the detection threshold, which in turn can be influenced by the linear system dynamics. Based on the findings presented, it is possible to conduct more focused experiments to confirm and add to these initial results.

Since 2019, the FAA and the EASA impose to all airline pilots to follow, as part of their training, a stall and recovery training. To mitigate the risks, this training takes place in ground-based simulators. To make it happen, models of the aircraft behaviour at high angle of attack need to be developed. Among the different modeling approaches possible, when available in such flight conditions (i.e. at high angles of attack, near and beyond stall), the use of actual flight test data is particularly interesting as it makes this work not being just a simulation study. Additionally, the method already proved to be efficient and effective for aerodynamic modeling purpose, especially in the nominal flight envelope, but it has also demonstrated interesting and encouraging results in modeling an extended (longitudinal) flight envelope, including stall entry up to post-stall conditions, using Kirchhoff theory of flow separation. With this work, it is
intended to support the idea that despite the development of digital techniques (CFD, physical simulation, etc.) flight data based models are still tools of major importance in order to create and develop models capable of positive transfer of training for the pilots in simulators. The purpose of this research is to try to fill a gap in the modeling of the lateral-directional dynamics of stalls using flight test data, by making use of an extended and adapted version of Kirchhoff Theory’s of flow separation. From a numerical point of view, it then seems that this approach
allows to improve by a few % the accuracy of the lateral-directional model.

Higher levels of automation in driving may allow drivers to engage in other activities, but may also increase the likelihood of Motion Sickness (MS). The exact causes of MS are not well understood, and various susceptibility factors(e.g. age, gender, ethnicity) can cause large individual differences. To better understand and predict MS, it is ideally studied in a safe and controlled environment, such as a driving simulator. However, the validity of driving simulator studies on MS as a proxy for on-road studies with real vehicles has not been properly evaluated. We conducted an experiment where the temporal aspects and symptom profiles of MS in a real-road driving scenario are compared to Simulator Sickness (SS) in a reproduction of this scenario in a motion-base driving simulator. A cohort of 25 participants was exposed to both the car and the simulator conditions. The scenarios consisted of sections of provocative(slaloming, stop-and-go) and normal driving. Sickening stimuli of the simulator were similar to the car accelerations in design (r = 0.51) but different in outcome (r = 0.27) as a result of motion cueing. MIsery Scale (MISC) scores on a 30 s interval, post-experiment Motion Sickness Assessment Questionnaire (MSAQ) scores, Galvanic Skin Response(GSR) and Electrogastrography (EGG) data were collected. Results showed significant correlations between the car and simulator conditions for 3 out of 4 MSAQ symptom categories (0.48 < r < 0.73, p < 0.02) and a relation (r = 0.57,p = 0.004) for individual sensitivity to sickness. Sickness onset times did not differ between the car and the simulator[F(1,308) = 4.80, p = 0.029], after Bonferroni corrections had been applied. Both MS and SS increased and decreased as a result of the driving style, with the effect being larger in the car condition, than in the simulator (for MISC [F(1,248)= 19.15, p = 0.000] and for GSR [F(1,230) = 5.55, p = 0.019]). Results from all four measures indicate that the severity of sickness was higher in the car as compared to in the simulator. EEG responses did not fully show expected outcomes. However, the signal quality was limited and dedicated EGG equipment may yield different results. Because individual sensitivity and temporal aspects of SS and MS were similar between the car and simulator but different in magnitude, we conclude relative validity for the simulator. As the human vestibular system, a prominent contributor in causing sickness, is solely sensitive to accelerations, we attribute the difference in magnitude due to downscaling of the vehicle motion in the simulator. In order to obtain absolute validity, either extensive training or considerable technological advances may be necessary.