Pilot manual control behavior has traditionally been modeled using purely visually driven quasi-linear frameworks. The role of proprioception and manipulator feedback (force and position) in shaping pilot dynamics for equalization has been theorized in literature, however well-grounded models that check all boxes remain elusive. This project investigated the feasibility of proprioceptive equalization utilizing a mechanical model of the neuromuscular system (NMS). Frequency domain analysis and root locus methods reveal that muscle spindle and tendon feedback may affect characteristics near the region of crossover, but cannot independently achieve the integrator-like open-loop characteristics required for effective equalization. A mixed equalization strategy based on combining force feedback with visual compensation is shown to be physiologically plausible and theoretically effective in reducing pilot effort. Qualitative comparison against experimental trends supports the conclusion that proprioception contributes to neuromuscular stabilization and performance enhancement; however, it could not conclusively prove the possibility of equalization.

Detecting human operator distraction is critical for safety in high-risk domains like aeronautics and the automotive industry. This paper explores distraction detection by framing the problem as both binary classification and anomaly detection and investigates the feasibility of training solely on not distracted data. Three approaches are evaluated: first, an ensemble Time Series Classifier for binary classification using both distracted and not distracted data; second, an LSTM-based auto-encoder trained only on not distracted data for binary classification; and third, a hybrid anomaly detector combining One-Class Support Vector Machine and Mahalanobis Distance, trained on not distracted data. These methods are tested on an existing dataset of pursuit tracking runs with a side task. The Time Series Classifier achieves 86.47% accuracy and an F1 score of 0.42 in the binary classification task, improving accuracy by over 30% and F1 by 0.17 compared to prior methods. However, it suffers from high false alarms and sensitivity to domain shifts due to its reliance on distracted and not distracted training data. The auto-encoder achieves an Area Under Curve of 0.76 and performs best in low false-positive scenarios. For the anomaly detection task, the combined anomaly detector outperforms individual models by reducing false alarms and improving recall, particularly in low false alarm regions, achieving precision above 0.85 and F1 scores exceeding 0.51, surpassing the ensemble time series classifier and the auto-encoder. These findings highlight the strengths of statistical methods and machine learning in distraction detection, providing a foundation for future advancements in human operator safety and distraction mitigation.

The identification of time-varying human operator (HO) dynamics is critical for advancing adaptive support systems in manual control tasks. This study evaluates the performance of Adaptive Model Selection (AMS), a framework extending recursive ARX identification methods, for estimating time-varying HO parameters, particularly effective time delay (τ(t)). Two configurations of AMS were tested: AMS_q, employing the z-domain ARX representation, and AMS_δ, utilizing the delta-domain ARX representation.

A Monte Carlo simulation framework was used to simulate a compensatory manual control task under varying conditions, including remnant noise (P_n) and dynamic transitions in system parameters. Stability and convergence rates of delay estimates were analyzed for different window sizes (W_s). Results show that the correlation between W_s and convergence time was linear and remained unaffected by remnant noise, demonstrating that window size is the primary determinant of responsiveness. Larger Ws improved stability but introduced tracking delays, whereas smaller W_s allowed for faster adaptation to dynamic changes at the cost of increased sensitivity to noise.

The comparative analysis between the configurations revealed a strong dependence of delay estimation accuracy on the precision of the natural frequency estimate of the neuromuscular system (NMS). The natural frequency estimation directly influences HO dynamic response modeling, and inaccuracies in this parameter propagate through the recursive identification process, affecting the reliability of delay estimates.

These findings underscore the critical role of window size and natural frequency estimation in determining the accuracy and stability of effective time delay estimation through AMS. This study provides a foundation for refining AMS to better balance stability and responsiveness in estimating time-varying HO dynamics. Such advancements can facilitate more accurate modeling of time-varying HO behavior, deepen understanding of HO adaptation, and contribute to the advancement of adaptive support systems.

Air traffic control (ATC) is transitioning towards a more automated system where human air traffic control officers (ATCOs) are increasingly supported by systems working at a high(er) level of automation (LOA). Made possible by advancements in computing power, artificial intelligence and a more data-driven air traffic management (ATM) system, automation is expected to address major issues, such as a global staff shortage, growing air traffic demand and environmental concerns.

On this shift towards greater reliance on automation, two main strategies can be identified that each have a distinct impact on the system's operators (i.e., ATCOs). Chapter 2 details how these differ between a traditional function-based strategy, where all flights are controlled at a gradually increasing LOA, and a constraint-based strategy, where a subset of flights is operated at a higher LOA than other flights. The former strategy brings many human-automation issues that have been widely demonstrated through empirical research, such as 'out-of-the-loop' situation awareness, transient workload peaks, skill erosion, boredom and reduced job satisfaction. The latter strategy has the advantage of avoiding mixed authority over individual flights by creating a more parallel system than the function-based serial system. The resulting human-autonomy team (HAT) accelerates the introduction of higher LOA in operational environments, fostering innovation.

The HAT perspective has only recently appeared on the radar of the ATC community, and practical examples of its potential and implications are scarce. An interesting example is found at Maastricht Upper Area Control Centre (MUAC), an air navigation service provider (ANSP) responsible for air traffic above 24,500 ft over Belgium, Luxembourg, the Netherlands, and part of Germany. MUAC is currently employing a constraint-based strategy in the development of a future shared airspace where ATC services for low-complexity routine flights are fully automated while complex flights stay with the ATCO. A key challenge for such an ATC system is to determine which flights should be allocated to either the human ATCO or the automation.

This research set out to broaden the knowledge about constraint-based automation in ATC and the desired allocation of flights in particular. Each chapter addresses a subquestion, often through empirical research with professional MUAC ATCOs. The research had three phases, starting with a first exploration, followed by an impact analysis of flight allocation on ATCO workflows and the role of flight complexity in this. The thesis concludes with a validation exercise consolidating all insights from the preceding chapters.

To test several preconditions and general ATCO acceptance of this novel concept, Chapter 3 begins with an exploratory simulator experiment. The participating ATCOs had full control over which flights they would delegate to the automation. Although pre-defined suggestions were presented, the ATCOs mostly ignored these. This experiment demonstrated the potential for allocating selected flights to either human or automation in a single airspace, but also stressed the importance of using a clever algorithm to determine this allocation. Geographic sector-based allocation, with automation handling all traffic in one sector and the ATCO all traffic in another sector, was rejected by the majority of participating ATCOs. They preferred an interaction-based allocation, hinting at the need to establish a complexity-score for each single flight.

Diving deeper into the impact that flight allocation might have on the workflow of an ATCO, Chapter 4 focuses on the core ATCO tasks: conflict detection and resolution (CD&R). Following a literature study and on-the-job ATCO observations, cognition flowcharts were constructed for these two tasks. Through an experiment with simplified static traffic scenarios, in which ATCOs had to detect and resolve conflicts, the most cognitively demanding types of traffic situations were searched for, as a means to quantify the various cognitive paths that can be traversed in the flowcharts. This turned out to be challenging, as ATCOs, like other experts, make frequent use of shortcuts and parallel processing. The constructed flowcharts can, however, serve as a starting point for the design of more human-like CD&R algorithms, such as used in this thesis' experiments. Automation that performs tasks in similar fashion as an ATCO might increase operator acceptance. This chapter's results stressed the importance of understanding flight-centric complexity before the impact of flight allocation on workflows can be determined.

To increase this understanding, the experiment in Chapter 5 used actual traffic snapshots overlaid with a single flight of interest for which the ATCOs had to indicate their perceived complexity. This individual flight complexity was a unique approach, compared to existing literature that mainly considers sector-wide complexity. Despite individual differences, flights on either end of the complexity scale were reliably identified. These results indicate that a flight allocation scheme may not need to be fine-tuned towards individual ATCO preferences. In general, a flight's complexity appears to be mostly driven by (potential) spatiotemporal interactions with other flights.

Consolidating the insights from preceding chapters, Chapter 6 discusses the most realistic and extensive experiment of this thesis. It replicates the experiment from Chapter 3 while addressing many of that experiment's shortcomings. Lessons learned in the preceding chapters led to several improvements, such as an increase in automation capabilities and communication, and more informed allocation schemes than the pragmatic schemes from the first experiment. In a direct comparison between two distinct allocation schemes, it was found that an interaction-based scheme is subjectively preferred by ATCOs and shows small efficiency benefits over a simpler flow-based allocation. In addition, it was concluded that automation should be sufficiently equipped to issue the same instructions as ATCOs, and should have the same notion of constraints from letters of agreement, to create a common ground and reduce mixed conflicts.

In conclusion, this thesis has brought forward the knowledge about flight allocation in an airspace that is shared between a human ATCO and a computer system. It can serve as a starting point for future research and development of highly automated ATC systems. Fully autonomous ATC will not become a reality in the short-term, but results show promising effects and a general feasibility of higher LOA applied to a constrained environment (i.e., a subset of flights). Researchers and ANSPs are encouraged to step beyond purely function-based visions on automation allocation and embrace a constraint-based automation strategy. This thesis has shown that a combination of these two strategies may lead to desired human-automation teamwork.

Recurrent training is crucial to helping pilots maintain or upgrade their flying skills. However, constantly monitoring and scrutinising pilots during training can lead to performance pressure. If pilots who are under pressure feel that their behaviour is affected it may ultimately reduce the effectiveness of recurrent training. To study the effects of performance pressure on pilot behaviour and learning, pilots (n=17) holding an airline transport pilot licence and type rating for large multi engine aircraft were invited to participate in a simulator experiment where they would learn a new autopilot system. The low-pressure group flew a series of training scenarios under no performance pressure whereas the high-pressure group flew the same training scenarios with prompts designed to induce performance pressure and raise their anxiety. It was hypothesised that the high-pressure group would be less incentivised to explore different functions of the autopilot due to the performance pressure, therefore leading to a lower quality of training which would be reflected in poorer knowledge of the autopilot system. Ultimately, whilst the high-pressure group did in fact undergo training with a significantly higher state anxiety. This was found to have no significant effect on their behaviour during training or their performance in a test scenario. The findings also suggest that the experiment would benefit from being repeated with more guidance given during training and more constrained scenarios to reduce the variation in the data.

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.

The introduction of Trajectory Based Operations (TBO) is set to change the operation on all levels of Air Traffic Control (ATC), including aerodrome control. Here, adherence to a planned four-dimensional trajectory is important to achieve a stable operation. At the same time, this concept provides possibilities for support tools to be used in ATC, by making use of new data and information. This research presents the Take-off Timing Support Tool (TTST) for aerodrome control, and more specifically runway control, at Amsterdam Airport Schiphol. It is incorporated into the already existing Electronic Flight Strip System and features a time axis on which flight strips can be placed. By dragging departures along the time axis a planning and sequence can be constructed, while taking into account the provided solution space to prevent conflicts. An experiment conducted with the TTST showed that professional air traffic controllers used the information from the tool in their decision-making, leading to a decrease in conflict count, and it was demonstrated that a stable planning could be established. Therefore, the TTST is a suitable platform to support TBO in aerodrome control and assists air traffic controllers with time-based control. No significant difference in departure interval was present. The workload did increase moderately.

Haptic shared controllers (HSCs) are a promising solution to prevent human over-reliance on automation during tasks such as car driving. However, research has shown that if the HSC is tuned incorrectly, then there is a risk of haptic conflicts between the human and HSC. To address this challenge, this paper presents the design and implementation of a novel adaptive HSC that continuously adjusts its look-ahead time. By estimating the time shift between the reference state of the HSC and the actual state, the HSC adapts to the look-ahead time of the human it is interacting with. Results from a human-in- the-loop experiment show that the novel HSC achieves similar subjective ratings as a fixed preview HSC, as well as a significant improvement over a fixed pursuit HSC. Going from pursuit to preview, objective experiment data shows that as the adaptive HSC adjusts its look-ahead time, haptic conflicts are reduced and tracking performance is increased. The presented findings are a step forward in designing haptic support systems with high chances of user acceptance. The proposed adaptive look-ahead algorithm provides a new method for online estimation of human look-ahead time, with or without a HSC in-the-loop.

Roll reversal errors, where the pilot tries to steer the aircraft back to wings-level but unintentionally increases the bank angle instead, have contributed to several accidents. Previous studies have shown that these errors can be caused by misinterpreting the attitude indicator (AI), with the figure-ground relations cited as contributing to this misinterpretation. A modified AI was developed, which uses several static monocular depth cues (color gradient, linear perspective lines, and shadow-light relationship) to strengthen the figure-ground relationship.
The modified version of the AI was compared to a baseline AI in a two-part flight simulator experiment where pilot reaction time and error rate, severity, and duration were measured. The first part induced the leans illusion making use of physiological adaptation to roll angle, distraction, and surprise. The second part simulated the leans illusion by simply rolling the simulator to the left or right. A group of 25 experienced commercial airline pilots performed a roll-to-level task in a moving-base simulator, which also provided spatially disorienting motion cues, using both the baseline and modified versions of the AI. While the modified
display had a lower error rate in the motion-opposite scenario when using the novel method (4.91% compared to 6.07%), no significant difference was found between the error rate of the two displays. The only significant difference was found in the reaction time, where the modified AI caused an increase in reaction time. The error rates and reaction times of the first part of the experiment did not match previous research. The novel disorientation method seemed to work best in a surprise scenario. While no significant differences were found between the modified AI and the baseline AI, it is still recommended to continue testing the modified AI with a new experiment setup, especially analyzing its effect in more extreme attitudes.

An important bottleneck in managing inbound traffic capacity is the limited size of the Terminal Manoeuvring Area, which should not become crowded by aircraft arriving from the Control Area. This is done by arrival metering, a process entailing sending aircraft through the Initial Approach Fix (IAF) at pre-established Expected Approach Times (EAT). Currently, area controllers are required to deliver aircraft at the IAF at their EAT ± 120 seconds. Reducing this window in the near future would allow for less buffer times between aircraft to increase capacity, potentially allowing implementation of time-based separation at the IAF as an initial step towards trajectory based operations. This research aims to produce a visual decision support tool to aid the aircraft sequencing process and achieving EAT adherence at the IAF, that is compatible with the currently in use or upcoming systems used in the Amsterdam Area Control Center. To avoid significant changes to the modus operandi of controllers, the aim is to show the user the operational constraints and focus on expediting traffic. An experiment was conducted with eight professional controllers to compare the newly developed tool with a rendition of the current interface. The results are positive, indicating less deviation from the EATs in all participants and overall lower participant workload as a result of using the proposed decision support tool.

Research into analysing the feasibility of using VR headsets to train helicopter pilots is in its infancy. While enough is known about what effects VR headsets have on pilots, a research gap exists in the field of explaining what exactly causes these effects to occur. This thesis aims to develop a visual model that can be used to analytically predict how pilot behaviour will be affected by a changing Field-of-View and explain why these changes occur. The scope of this model is analysing where in the pilot’s vision, the point of most feedback lies, and what happens when this point gets clipped out of view while the pilot controls a two degree-of-freedom helicopter model. Subsequently, human-in-the-loop experiments were carried out where the participants had to do velocity disturbance rejection tasks and pitch disturbance rejection tasks for various fields-of-view, to try and validate the model. It was predicted that the performance for the pitch disturbance rejection task would not be affected by the field-of-view but the performance for the velocity disturbance rejection task would degrade as field-of-view was lowered. Overall, there was no significant trend that emerged from the experiments for either task, but significant insights were gained into the scope of the visual model. Based on these, and other findings, the model did a good job in identifying why performance changes and in setting up future researchers for success in this field.

Predictive aircraft maintenance is a maintenance strategy that aims to reduce the number of failures, the number of inspections, the number of maintenance tasks and the aircraft maintenance costs. Aircraft are equipped with health monitoring systems, where sensors continuously measure the condition of the aircraft components. In predictive maintenance, these sensor measurements are used to estimate the time left until the failure of these components, called the Remaining Useful Life (RUL). These RUL prognostics are subsequently used to optimize the aircraft maintenance schedule. There are several challenges that complicate the implementation of predictive aircraft maintenance in practice. In this thesis, the threemain challenges are addressed.

The aviation industry's reliance on automation raises concerns about pilot complacency, necessitating continuous pilot proficiency measures. To that end, real-time pilot skill feedback is vital—through alerts on declining skill levels or scalable levels of autonomy. Current cybernetic methods are limited as they assume linearity and time-invariance of human behavior and lack real-time capability. Neural Networks (NNs) offer a solution but face challenges such as high computational costs and limited generalization capability. To overcome these issues, this paper introduces a new and compact Residual Network for eXplainable Convolutional MTS Classification (RN-XCM) designed explicitly to classify pilot skill levels. Results demonstrate RN-XCM's ability to accurately classify skill levels based on 1.2 seconds of dual-axis control data, achieving a test accuracy of up to 93.50%, while requiring 50% less training time than competing NN models. It also achieves a test accuracy of 80.16% for previously unseen subjects, signifying its competence as a one-size-fits-all classifier. Notably, RN-XCM performs 17.88% better when classifying dual-axis tracking tasks over single-axis tracking tasks. Overall, the possibility of real-time feedback provided by the RN-XCM can enable quantitative evaluation of pilot control behavior, therefore enhancing safety and facilitating smoother interactions between pilots and aircraft.

Autonomous vehicles are becoming increasingly prevalent in society, but the transition from active driver to passive passenger is known to increase the risk and severity of motion sickness. Motion anticipation is a critical factor in this difference, and visual information is known to be a major contributor to motion anticipation. However, the underlying mechanisms of how visual information influences motion sickness remain largely unknown. This study aims to investigate the effect of visual translation on motion sickness development. To accomplish this, a proposal was made to extend an existing motion sickness prediction model with visual translational components. A human-out-of-the-loop experiment was conducted in a moving-base simulator to test the proposed model's reliability and determine whether visual translation influences motion sickness. The experiment simulated an urban ride that featured repetitive accelerating and breaking maneuvers on a straight road. Eighteen participants were subjected to three different visual conditions that varied in the amount of global optical flow, while the simulator's motion was identical. The results suggest that congruent visual translation may lead to a minor, non-significant reduction of motion sickness, while the amount of optical flow does not seem to affect it. The proposed motion sickness model was able to predict a slight decrease in motion sickness for conditions with visual optical flow, as demonstrated by the experiment. Nonetheless, it is still debatable whether congruent visual translation may influence the onset and severity of motion sickness, as the results of this study could not confirm this.

The joint management of airspace by human controllers and automated agents is gaining prevalence in nextgeneration Air Traffic Control (ATC). In such settings, human controllers are challenged with maintaining situation awareness while dealing with intricate and often opaque automated technologies. This study probed the potential of a concept interface, which augments traditional radar displays, to facilitate human controller-automation teamwork. With visual explanations and interactive elements, it intended to make the control behavior of automated agents transparent and comprehensible to human operators. The primary objective was to explore whether features from a non-conventional design could offer insights for designing interfaces for highly automated ATC operations. The concept was tested in a prototype-based user experience evaluation with static traffic scenarios, involving 15 active air traffic controllers from the Maastricht Upper Area Control Center. Results indicated that the visual explanations of the interface enabled controllers to comprehend the plans and reasoning underlying automation’s behavior. A significant correlation between controllers’ evaluations and their attitudes toward novel technologies underlines the influence of pre-existing beliefs about automation on the acceptance of new interface tools. Future research is warranted to streamline parts of the design and further investigate the interface’s capacity for explaining automation’s actions in dynamic, real-time scenarios.

This study investigated the effect of manual and autopilot control on hazard and failure detection in aviation, in cases where the autopilot can masks or diminishes cues of aircraft movement. This mechanism could result in loss of situation awareness, upsets and ultimately accidents. Twenty airline pilots participated in an experiment in which two scenarios, an engine failure and icing accumulation hazard, were flown in both manual and autopilot control conditions. Results show no significant difference in detection time for the engine failure scenario between the two modes, but a marginally significant difference in the icing accumulation scenario and a clear difference in the way pilots detected the failure/hazard. Manual control may provide clearer cues, but automation may lower workload and allow for better monitoring. Monitoring of the flight controls was found to be an important factor in hazard detection, with pilots who had their hands resting on the controls having significantly lower detection times. The study provides insights on the importance of active monitoring of flight controls and on the impact of non-backdriven controls. The findings suggest that understanding the advantages and disadvantages of autopilot use is crucial for training pilots effectively.

This study explored several anticipatory interfaces as motion sickness mitigation tool in autonomously driving vehicles. The interfaces were designed to lower the mismatch between the expected motion and the sensory input in a moving body, by providing anticipatory information on upcoming vehicle motion. The two proposed interfaces were a Light only interface and a Combined interface featuring both light and sound. The interfaces present information on the direction and magnitude proportional to the upcoming vehicle accelerations, 0.9s before the actual event. The interfaces were tested in separate trials and compared to a Baseline trial. The comparison was done by a mixed-order, within-participants experiment (n=18), performed in the SIMONA Research Simulator, with identical 30-minute driving conditions. To test the interfaces’ effectiveness in the most motion sickness-triggering setting, subjects got a reading task on a handheld tablet. The results were quantified on an 11-point motion sickness scale and subjective survey questions. Improvements in motion sickness and subjective experience were found with both interfaces. However, only the Combined interface trial showed significant improvements compare to the Baseline. Neither of the interfaces was considered intrusive, while only the Combined interface was considered comfortable. The survey also clearly favored the Combined interface in scores on timing cor- rectness, intuitiveness, and subjective cue correctness. Although the Combined interface had the best results, strong individual preferences were reported for both interfaces, a hypothetical sound interface, or no interface. Further research is advised to focus on the Combined interface while considering optional light and/or sound disengagement in line with personal preference.

The Dutch Air Traffic Control has implemented fixed approach trajectories within approach control during night operations when traffic density is low. During the day, when traffic density is higher, vector-based operations are used. The Dutch Air Traffic Control aims to implement fixed approach trajectories when traffic density is high. This shift from the current vector-based operations will allow aircraft to perform a continuous descent and fly around inhabited areas, reducing the emissions and noise. To improve the capacity of the landing runway in strong headwind conditions, time-based separation is envisioned. Separating aircraft on time will result in a constant runway capacity in all wind conditions. However, to make these changes possible, additional decision support tools are required to assist controllers and reduce their workload. This paper discusses the use of a new toolset, centered around a time-space diagram which shows the expected arrival time and distance-to-go to a selected reference waypoint. Results of an experiment with eight professional air traffic controllers show that the added tools allowed future conflicts to be solved sooner and with fewer instructions. Aircraft followed the fixed approach trajectories better in the latter stages of the approach when compared to current vector-based operations. Although a slight increase in runway capacity was observed with the new toolset, it was not statistically significant. The workload of the controllers did not show a difference when using the new toolset. In conclusion, the added tools enabled the controllers to combine fixed approach trajectories with time-based separation. However, to reduce perceived workload, future designs should aim to integrate the time-space diagram on the main radar screen such that controllers do not need to switch attention between screens.