The interaction between humans and AI in safety-critical systems presents a unique set of challenges that remain partially addressed by existing frameworks. These challenges stem from the complex interplay of requirements for transparency, trust, and explainability, coupled with the necessity for robust and safe decision-making. A framework that holistically integrates human and AI capabilities while addressing these concerns is notably required, bridging the critical gaps in designing, deploying, and maintaining safe and effective systems. This paper proposes a holistic conceptual framework for critical infrastructures by adopting an interdisciplinary approach. It integrates traditionally distinct fields such as mathematics, decision theory, computer science, philosophy, psychology, and cognitive engineering and draws on specialized engineering domains, particularly energy, mobility, and aeronautics. Its flexibility is further demonstrated through a case study on power grid management.

Air traffic control is advancing digitalization by developing advanced decision-support systems, where the way information is presented to operators plays a central role in shaping performance. However, the effects of different visual representations within these systems on human decision-making remain not fully understood. In this study, we compared two Conflict Detection and Resolution (CD&R) tools: the Highly Interactive Problem Solver (HIPS) and the Solution Space Diagram (SSD). Although both systems are grounded in the same control problem, they differ in how they represent the control constraints that define conflict conditions and feasible responses. Through a human-in-the-loop experiment under low-and high-traffic conditions, we analyzed how these differences influence decision-making. Results showed that, particularly in low-density traffic, HIPS enabled quicker responses, fewer commands, and smaller safety margins, whereas SSD, despite receiving more favorable subjective ratings, led to greater variability in actions. These findings suggest that visualization significantly impacts decision-making consistency and efficiency. However, in highly complex environments, overall effectiveness may depend more on operators’ ability to shift and adapt decision-making patterns facilitated by the interface than on specific visual elements.

Computer algorithms facilitate increased automation in various human-centered work areas to improve operational safety and efficiency. Algorithmic transparency is considered essential for human operators, policy makers and system developers, as it allows them to understand the capabilities and limitations of an algorithm. In this research, we focus on path-planning algorithms and propose a purely visual approach to achieve their transparency. This approach extracts and portrays information directly from the algorithms, aiming to visually reveal their inner workings. Benchmark tests indicate that extracting information from path-planning algorithms may significantly slow them down. For time-constrained operations, it is recommended to store only the necessary data during the pathfinding process and perform information extraction afterwards. Based on theories from cognitive engineering, six transparency levels were designed to chunk meaningful information pertaining path-planning algorithms. A user study among non-experts ( N = 40 ) was then conducted to evaluate the impact of visual algorithmic transparency on human understanding. The results suggest that increased transparency levels allow non-experts to more correctly and confidently understand the details of a path-planning algorithm. However, it is also found that certain transparency levels can lead to confusion, especially when the algorithm behaves in a way contrary to human expectations. This study further reveals that, given the same level of transparency, sampling-based algorithms may be easier to comprehend than graph-based algorithms. This research can serve as a reference for how to achieve transparency in path-planning-related applications and how to hierarchically portray and organize transparency information.

Air traffic controller workload is a limiting factor in the current air traffic management system. Adaptive support systems have the potential to balance controller workload and gain acceptance as they provide support during times of need. Challenges in the design of adaptive support systems are to decide when and how to trigger support. The goal of this study is to gain empirical insights into these challenges through a human-in-the-loop experiment, featuring a simplified air traffic control environment in which a novel triggering mechanism uses the quality of the controller's decisions to determine when support is needed. The designed system seeks to prevent high workload conditions by providing resolution advisories when the controller exceeds a threshold of “self-complicating” decisions. Results indicate that the new system is indeed capable of increasing the efficiency and safety compared to full manual control without intervention. More adaptive support, however, increased the frustration of participants, decreased acceptance, and did not result in improved workload ratings. These findings suggest that, unless we can better infer human intent in complex work environments, adaptive support at the level of decision-making is problematic. A potentially more fruitful direction is to provide support at the level of information integration, with full decision-making authority with the human.

In the near future, air traffic controllers are expected to adhere to stringent time and position constraints in controlling traffic. For this new task, new decision-support tools are required which include higher levels of automation, whilst letting humans remain to be the ultimate responsible for the safety of operations. In previous research, an advanced human-machine interface was designed and evaluated that allows controllers to manipulate four-dimensional flight plans of each individual aircraft. In this research, a higher level of automation is explored by designing a new interface prototype that enables controllers to manipulate multiple flows of traffic by facilitating interaction with a path-planning algorithm. A first evaluation of this interface, in which five participants were asked to structure a perturbed airspace as they saw fit, showed that the participants were able to influence the algorithm as they desired and were supported by the interface that visualized the inner workings of the algorithm. However, human influence did not improve the solutions in terms of sector robustness and efficiency, as compared the previously designed interface for aircraft. Therefore, improvements and its use case warrant further research.

Future Air Traffic Management concepts will require air traffic controllers to move from a tactical to a strategic way of operation. This paper evaluates two novel concepts which support controllers to perform four-dimensional trajectory management in a contingency situation. The Travel Space Representation and Time-Space Diagram are both solution-space based displays where automation calculates all possible actions in real-time. All decision-making is still to be done by the operator but is greatly facilitated by this automation, as it shows all possible actions at a glance. An experiment is described which evaluated the performance of novice controllers in managing a sector where suddenly a bad weather cell emerged, requiring them to re-route traffic in space and time. Results show that the display concepts work well and support operators even in complex situations with a heterogeneous mix of aircraft types and speeds. Performance and workload indicators become worse for the higher-density, higher-heterogeneity situations.

Future Air Traffic Management is expected to shift towards four dimensional trajectory (4DT) management, requiring new decision support tools for air traffic controllers to meet stringent time and position constraints. In previous work, a prototype human-machine interface has been developed for 4D trajectory manipulations of single aircraft. This paper describes a tool for multi-aircraft manipulation and investigates its potential control efficiency benefits. A human-in-the-loop experiment (N = 13) has been conducted using scenarios with sector disruptions and varying conflict geometry. Results show that participants preferred to use multi-aircraft manipulation for groups of aircraft having small convergence angles. Since the current implementation involves re-routing all selected aircraft through one common waypoint (referred to as a 'merge point'), extra additional track miles were flown and airspace robustness reduces. Regarding efficiency and safety, multi-aircraft trajectory manipulation seems favourable only for smaller convergence angles, although this also depends on the way the operators place the aircraft merging points. For future development, attention should be devoted to making flight efficiency constraints of each aircraft more salient, enabling controllers to better time the rerouting multiple aircraft and more fairly distribute re-routing costs.

In many work domains, the push toward higher levels of automation raises the concern of diminishing human expertise. Ecological interfaces could help operators in retaining and potentially even in acquiring expertise as they are hypothesized to lead to a deeper understanding of the work domain. This study explores the short-term impact of ecological interfaces on knowledge development and compares the results with an instruction-based training method. To monitor and compare students' progress, their decision-making strategies, identified from verbal comments recorded in “think-aloud” simulator sessions, are mapped onto the decision ladder. This method has been applied to an experiment (N = 16) aimed at training novices in conflict detection and resolution (CD&R) within a simplified air traffic control context. Results show that the overall CD&R performance in the final measurement sessions, featuring a transfer manipulation, was not significantly different between the “ecological” and “instructional” groups. In terms of cognitive behavior, however, students in the ecological group exhibited more laborious rule and knowledge-based behaviors that sparked goal-oriented thoughts and corresponding control performances beyond the CD&R task. These findings indicate that ecological interfaces can change how people think and approach a control problem, even after removing the support. It is therefore reasonable to believe that ecological interfaces can play an important role in the early stages of deep knowledge development.

Perspective flight-path displays are a viable alternative for the aircraft primary flight display, but increases the pilot head-down time. A haptic interface is developed to counter this effect and increase the task-sharing performance during approach. An experiment (n=12) was conducted to test the effects of the haptic design on primary task performance with a tunnel-in-the-sky display, in a low and high workload condition. To investigate the effects of the haptic interface on the headdown time, a secondary task was presented on the simulator outside visual, in the form of bucket-shaped figures, requiring participants to indicate the direction of the one divergent figure. Secondary task performance was measured by success rate, average time to answer correctly and – by means of eye-tracker measurements – head-up time and number of gaze switches. Pilots also provided a subjective measure of their mental effort after each run. Results show that haptic feedback significantly increases both primary and secondary task performance of the pilots, especially when the primary task is more challenging. Workload ratings are significantly lower, and head-up time increases with haptic feedback.

Haptic interfaces have been developed to assist human operators controlling unmanned aerial vehicles (UAV) beyond line-of-sight. These systems complement the predominantly camera-based visual interfaces and act like a collision avoidance system: when nearing obstacles, the operator gets re-directed by the haptic force feedback. Previous research showed that, although being successful in reducing the number of collisions, these haptic interfaces also lead to lower user acceptance, as the operators do not always understand the system's intentions. This paper discusses two novel visualizations which were developed to increase operator acceptance. Both designs were evaluated in a human-in-the-loop experiment (n = 12). Results from acceptance-related questionnaires show that our subjects preferred teleoperating the UAV with the visualizations active. Acceptance ratings were higher for the same levels of safety, performance and operator workload.

Ecological interface design (EID) was originally developed in the context of process control but has been extended into many domains where technology has resulted in both changing work demands and increased opportunities for improved interface applications. This paper gives an overview of the application of the EID to the control of vehicle locomotion, either from within the vehicle, as a driver or a pilot, or from the outside, as an operator or a (air traffic) controller. It discusses lessons learned from the application of the EID for the vehicle locomotion control task and focuses on how the methodology can be applied to this domain. Specific issues identified are that the planning and control of a vehicle simultaneously spans multiple time scales and that the interface must be designed considering the format in which the control input is defined. Also, due to the extensive standardization of the instrumentation and training certification, changes introduced by the new displays must initially be additional to the existing displays. Chosen representations must also be shown in a format that matches the current instrumentation and the directly observable outside world.

The introduction of 4-D trajectory-based operations will require the development of new and more advanced “human-centered” decision support tools for future air traffic controllers. One approach to the design of human-centered decision aids is ecological interface design, which focuses on visualizing the boundaries of safe system performance rather than prescribing predetermined strategies or discrete solutions. Previous studies with ecological interfaces in the aviation domain revealed that humans sometimes opted for control actions close to these boundaries, giving rise to a general concern about the robustness of control actions. The goal of this study has been to empirically investigate how effectively an ecological interface for 4-D trajectory management, as developed in a previous study, supports the preservation of airspace robustness. For this purpose, a metric has been developed to evaluate both minimum and average sector-based and control-based robustness. Special attention was paid to quantifying and measuring the effect of expertise level on the robustness of human-generated control actions. Results of a human-in-the-loop experiment indicate that expert participants were most robust in their control actions, as compared with either skilled or novice participants. This result suggests that boundary-seeking control actions with ecological interfaces are mainly dependent on the level of expertise and the control strategies of the end user.

Cognitive engineering researchers have long studied the complexity and reliability of human–automation interaction. Historically, though, the area of human–automation decision-making compatibility has received less attention. Paradoxically, this could in the future become one of the most critical issues of all, as mismatches between human and automation problem-solving styles could threaten the adoption of automation. This paper presents the concept ofstrategic conformanceas a potential key factor influencing initial acceptance of automation, specifically decision aiding systems capable of guiding decision and action. Here, strategic conformance represents the match in problem-solving style between decision aiding automation and the individual operator. The theoretical foundation builds on the compatibility construct found in technology acceptance theories such as the innovation diffusion and technology acceptance models. The paper concludes with a critical discussion on the limitations and drawbacks of strategic conformance. It is proposed that the construct would be most applicable at the introductory phase of new decision aiding automation, in helping to foster operators’ initial acceptance of such automation.

The ecological interface design (EID) paradigm was introduced in the process control domain 25 years ago by Kim Vicente and Jens Rasmussen, as a way to help operators cope with system complexity and events unanticipated in the design of automated control systems. Since that time, this perspective has sparked interest in other safety-critical sociotechnical domains where humans cooperate with computerized systems to ensure safe and efficient system behavior. Many of our own, but also other explorations have, however, resulted in several usability concerns and misconceptions about the EID perspective as a viable design approach. This paper discusses some of these concerns and misconceptions, where the final goal is to get past the EID label and to consider the general lessons relative to the demands and opportunities that advanced information technologies offer and complex systems require. This paper concludes with a preliminary outlook for the future of EID, where it is anticipated that the adjective “ecological” will become increasingly redundant, as the focus on supporting “productive thinking” becomes the dominant paradigm for engineering representations.

The velocity obstacle (VO) representation, a mathematical representation that supports separation assurance for autonomous vehicle navigation, can provide improved performance when the intended trajectory (intent) of the involved vehicles is shared among each other. This paper proposes a fully analytical closed mathematical form for VOs as an intuitive and flexible calculation method, which is able to incorporate time-varying concepts such as trajectory prediction errors. Several study cases for specific traffic situations are explored, and possibilities to extend the method to the third dimension are discussed. Analysis showed that the calculation of the VOs with intent information could show self-curve intersections when maneuvers at close distance are required. Further elimination of these intersections might be necessary for certain specific applications.

To lower the workload and increase the productivity of air traffic controllers in future scenarios, decision-support systems that functions on a higher level of control authority would be desired. However, increasing the control authority of automation asks for some important considerations in terms of design to ensure effective human-machine coordination. One of the factors that is hypothesized to affect human-machine collaboration, and one that is often overlooked in the design of such systems, is the solution transparency offered by the automated system. That is, when an automated system reveals some of its inner workings, it is expected to promote better system understanding and an increased acceptance of resolution advisories offered by that system. To this end, this paper investigates the effects of automation transparency on the acceptance of automated resolution advisories in conflict detection and resolution task for air traffic controllers. An exploratory experiment, featuring twelve aerospace students, was performed to study these effects. The results indicated that no main effects of transparency on acceptance were found, but also that there was a significant interaction effect between conflict geometry and transparency. For future research, an experiment is proposed with professional air traffic controllers and a larger sample size to increase the power of the statistics.

The current evolution of the Air Traffic Management system towards trajectory-based operations is foreseen to bring large changes to the work domain of the Air Traffic Controller. Although this new form of Air Traffic Control leans heavily on the introduction of advanced automation, the general consensus is that the human must remain actively involved in the decision-making loop, and retain the ultimate responsibility for the safety of operations. These responsibilities, together with the complexities of the new task, require the development of innovative decision support tools. In previous research, and following the principles of Ecological Interface Design, a constraint-based decision support tool has been developed for the task of strategic trajectory manipulation. Rather than presenting discrete optimized solutions to the controller, this Travel Space Representation visualizes the constraints for safe control in the form of a set of `go' and `no go' areas. A validation experiment demonstrated that when using this tool, controllers sometimes opted for controlling close to the boundaries of safe control, or for resolutions in narrow control spaces. These results gave rise to concern that such a representation could actually work against the flexibility of the system to cope with inherent system variability. In this study, a metric and two measures have been developed in order to quantify and compare trajectory-based robustness to probabilistic disturbances. A batch-analysis has been performed to investigate how these measures vary for a crossing pair of aircraft under various geometry. Results show that the metric captures additional information which is currently not represented in the tool. When visualized to the controller, this could support them to choose more robust control strategies.

Air traffic control uses arrival managers (AMANs) to schedule an inbound stream of aircraft. Most state-of-the-art AMANs require the operator to analyse and integrate the schedule data to optimize the use of runways. None of the currently operational systems explicitly visualize inefficient use of runway capacity or the imbalance between available capacity and demand. Rather, these decisions are based on predicted arrival times and require an initial planning to determine potential delay in the active runway configuration. The research presented in this paper focuses on supporting the runway planning decision-making process using an experimental ecological AMAN by visualizing imbalances between capacity and demand. An indepth analysis of the arrival management process has been conducted and a concept interface is proposed. A human-in-the-loop experiment was conducted to investigate the effect of visualizing performance deviation information on the quality of runway planning. No significant differences in planning performance have been found between these two conditions, most likely due to the nature of the various traffic scenarios and the high overall workload during the experiment. However, more consistent performance has been observed when the performance deviation information was available to the participants and more improvements of the runway planning closer to landing were visible. Both trends are desired effects of visualizing performance deviation and warrant further research.

Changes in Air Traffic Management, exemplified by the NextGen and SESAR projects, intend to achieve ATM systems with a higher capacity and also higher efficiency, through more direct routing. At the same time, air traffic safety should continue to increase. An important step in this process can be the partial or full transfer of responsibility for separation to the flight deck, using Airborne Separation Assurance Systems (ASAS). In previous work, our group has developed flight deck displays that enable pilots to perform the task of self-separation through the visualization of conflicts on the displays. In various tests with these displays, pilots successfully performed the task of self-separation, using only the implicit coordination based on conflict geometry as presented on the display. In order to further investigate the emergent features of the use of such displays, a set of rules was created to mimic the behavior observed in the experiments. Using this simulation, a Monte Carlo experiment was run, in which the properties of encounters between two aircraft were further investigated. The simulations results indicate that the implicit coordination used in our set-up is not always successful. Furthermore, instructions for solving the conflict that emphasize quick resolution can be detrimental to safety.